diff --git a/paper-releases/pbisc-diffusion/updates/20261013T022035Z-pbisc-red-scsemi-blue/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/data/results_review_20261013_provenance.json b/paper-releases/pbisc-diffusion/updates/20261013T022035Z-pbisc-red-scsemi-blue/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/data/results_review_20261013_provenance.json new file mode 100644 index 0000000000000000000000000000000000000000..ae6e66514041f479369309cb47cbeea87dc55162 --- /dev/null +++ b/paper-releases/pbisc-diffusion/updates/20261013T022035Z-pbisc-red-scsemi-blue/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/data/results_review_20261013_provenance.json @@ -0,0 +1,2562 @@ +{ + "schema": "pbisc.results-review.reuse.v2", + "style_revision": 4, + "as_of_kst": "2026-10-13T11:17:36.769774+09:00", + "slides": [ + { + "kind": "title", + "title": "PBISC-Diffusion", + "subtitle": "Results review \u00b7 existing evidence and protocol updates \u00b7 13 October 2026" + }, + { + "kind": "agenda", + "title": "Six figures: evidence available now", + "items": [ + "Fig. 1 Pretraining, diffusion and bulk conditioning", + "Fig. 2 Model design and paired-data requirements", + "Fig. 3 Healthy reconstruction: quality, composition and time", + "Fig. 4 Disease signals and deconvolution benchmarks", + "Fig. 5 Rare cell states: corrected analysis of existing outputs", + "Fig. 6 Bulk-only reproduction: awaiting the new dataset" + ] + }, + { + "kind": "content", + "section": "plan", + "section_label": "Fig. 1a", + "text_source": { + "file": "data/figure_plan_slide_text.json", + "index": 8 + }, + "title": "PBISC-Diffusion: pretraining to generated cells", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/fig1b_overview_nature.png" + ], + "keywords": [], + "footnote": "Schematic. Conditioning recipe size and the number of generated cells are distinct; no cell-type quotas.", + "sources": "tools/figure_plan.py; manuscript figures/fig01_architecture_nature.svg", + "footnote_full": "Schematic. Conditioning recipe size and the number of generated cells are distinct; no cell-type quotas.\nOriginal context: Candidate replacement for the engineering diagram (kept in this section as details and for Methods). Source: paper-release-20260928/manuscript/figures/fig01_architecture_nature.svg (scripts/build_figure1_architecture_nature.py).", + "numbers": [], + "original_slide": 7, + "figures_before_palette": [ + "figures/fig1b_overview_nature.png" + ] + }, + { + "kind": "content", + "section": "1", + "section_label": "Fig. 1b", + "title": "Bulk conditioning", + "layout": "diagram", + "figures": [ + "figures/sec02_bulk_conditioning.png" + ], + "keywords": [ + "One shared bulk condition; independent cell draws", + "Expression, depth and measured-gene masks", + "No cell-type or disease labels as model input" + ], + "footnote": "Final 6k model. Recipe size K conditions the bulk; the number of generated cells is chosen separately.", + "sources": "tools/sec02_backbone.py; data/sec02_bulk_conditioning.csv, data/sec02_slide_text.csv", + "footnote_full": "Final 6k model. Recipe size K conditions the bulk; the number of generated cells is chosen separately.\nOriginal context: Final model f01 6k 1x, part 16. Bar heights schematic.", + "original_slide": 8, + "figures_before_palette": [ + "figures/sec02_bulk_conditioning.png" + ] + }, + { + "kind": "content", + "section": "plan", + "section_label": "Fig. 1c", + "text_source": { + "file": "data/figure_plan_slide_text.json", + "index": 9 + }, + "title": "Pretraining corpus", + "layout": "full", + "figures": [ + "figures/fig1d_corpus.png" + ], + "keywords": [ + "41.4 M cells, 22 datasets, 3,199 donors", + "88.5% healthy donors; 23.6% cultured in vitro", + "Labels describe the corpus; never model input" + ], + "footnote": "Corpus labels describe the data; they are not model inputs. Dataset composition is not sampler exposure.", + "sources": "tools/figure_plan.py; manuscript figures/fig01_corpus_nature.png, source_data/corpus/", + "footnote_full": "Corpus labels describe the data; they are not model inputs. Dataset composition is not sampler exposure.\nOriginal context: Source: paper-release-20260928/manuscript/source_data/corpus/ (build_corpus_tables.py) and figures/fig01_corpus_nature.png (build_figure1_corpus.py).", + "numbers": [ + { + "value": "41,395,184 cells; 22 datasets; 3,199 donors; healthy 88.5%; in vitro 23.6%", + "source": "manuscript/source_data/corpus/corpus_summary.json" + } + ], + "original_slide": 9, + "figures_before_palette": [ + "figures/fig1d_corpus.png" + ] + }, + { + "kind": "content", + "title": "Native-pseudobulk model comparisons", + "section_label": "Fig. 2a", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/review_style_20261013_ablation.png" + ], + "keywords": [], + "footnote": "Completed held-out rescoring. Architecture: 1% detection floor; gene-panel comparison: 0.1% floor.", + "footnote_full": "Completed held-out rescoring. Architecture: 1% detection floor; gene-panel comparison: 0.1% floor.", + "sources": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_ablation.pdf", + "figures_before_palette": [ + "figures/review_style_20261013_ablation.png" + ] + }, + { + "kind": "content", + "section": "1", + "section_label": "Fig. 2b", + "title": "Gene panels: performance and coverage", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/sec03_panel.png" + ], + "keywords": [ + "6k \u2248 3k: 3k r, DEG inside 6k seed range", + "2k best L1; 15k worst r, L1", + "6k kept: first panel with CLEC9A (cDC1)" + ], + "footnote": "Existing panel benchmark. 6k: five sampling seeds; other panels: one seed. Different axis from the architecture sweep.", + "say": "Runs f01_{2k,3k,6k,p15k}_1x, part 16, Bernoulli, EMA, DDIM 50. Different gene axis from the previous slide, so the two slides cannot be compared. Loss curves are divided by each panel's own part-16 value.", + "sources": "tools/sec03_panel.py; data/sec03_panel_bench.csv, data/sec03_panel_seeds.csv, data/sec03_panel_loss.csv, data/sec03_panel_headline.csv", + "original_slide": 12, + "footnote_full": "Existing panel benchmark. 6k: five sampling seeds; other panels: one seed. Different axis from the architecture sweep.\nOriginal context: Existing panel benchmark. 6k: five sampling seeds; other panels: one seed. Different axis from the architecture sweep.", + "figures_before_palette": [ + "figures/sec03_panel.png" + ] + }, + { + "kind": "content", + "title": "Healthy bulk: full fine-tuning and LoRA rank", + "section_label": "Fig. 2c", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/review_style_20261013_rank.png" + ], + "keywords": [], + "footnote": "Original donor-level five-fold comparison. LoRA rank 16 is a quality/parameter compromise.", + "footnote_full": "Original donor-level five-fold comparison. LoRA rank 16 is a quality/parameter compromise.", + "sources": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_rank.pdf", + "figures_before_palette": [ + "figures/review_style_20261013_rank.png" + ] + }, + { + "kind": "content", + "section": "3", + "section_label": "Fig. 2d", + "text_source": { + "file": "data/fig23_healthy_slide_text.json", + "index": 6 + }, + "title": "Historical paired-data learning curve", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/fig3_healthy_pairs.png" + ], + "keywords": [ + "Coverage at every n: LoRA 0.76\u20130.84, scSemiProfiler 0.04", + "Composition L1 lower for LoRA at every n", + "5-fold: LoRA 0.16 vs 0.22 L1" + ], + "footnote": "LoRA: n training + 7 validation. scSemi: n + 7 representatives, with CV-HVG reuse. Not the strict total-n protocol.", + "sources": "tools/fig23_healthy.py; data/fig23_healthy.csv", + "refs": [ + "Wang 2024, Nat Commun" + ], + "refs_full": [ + "Wang et al. scSemiProfiler: advancing large-scale single-cell studies through semi-profiling with deep generative models and active learning. Nat Commun 15:5989 (2024). doi:10.1038/s41467-024-50150-1" + ], + "say": "Giving scSemiProfiler exactly LoRA's pairs, from five up to every training donor, does not close the gap in coverage or composition.", + "footnote_full": "LoRA: n training + 7 validation. scSemi: n + 7 representatives, with CV-HVG reuse. Not the strict total-n protocol.\nOriginal context: ImMicro, donor-level 5-fold CV pooled at each pair count: for every fold, scSemiProfiler's representatives are exactly the n pairs LoRA trained on (plus the inner-validation pairs); 'all' (open symbols, right of the dotted line) = every training donor of the fold, i.e. the 5-fold result, reported separately from the 5-pair comparison. Source: data/sec08_pairs_realism.csv, data/sec08_pairs_signal.csv, data/sec09_same_pairs.csv.", + "numbers": [ + { + "value": "pairs 5 comp_l1 scsemi 0.2184; pairs 10 comp_l1 scsemi 0.2269; pairs 20 comp_l1 scsemi 0.2232; pairs 40 comp_l1 scsemi 0.2213; pairs all comp_l1 scsemi 0.217; pairs 5 comp_l1 lora 0.1715; pairs 10 comp_l1 lora 0.1879; pairs 20 comp_l1 lora 0.1723; pairs 40 comp_l1 lora 0.1714; pairs all comp_l1 lora 0.1617; pairs 5 coverage scsemi 0.03819; pairs 10 coverage scsemi 0.03801; pairs 20 coverage scsemi 0.0372; pairs 40 coverage scsemi 0.03844; pairs all coverage scsemi 0.03844; pairs 5 coverage lora 0.8392; pairs 10 coverage lora 0.7857; pairs 20 coverage lora 0.7552; pairs 40 coverage lora 0.7784; pairs all coverage lora 0.7843; pairs 5 nk_fraction_r scsemi 0.2651; pairs 10 nk_fraction_r scsemi 0.33; pairs 20 nk_fraction_r scsemi 0.4762; pairs 40 nk_fraction_r scsemi 0.4937; pairs all nk_fraction_r scsemi 0.4245; pairs 5 nk_fraction_r lora 0.1993; pairs 10 nk_fraction_r lora 0.3972; pairs 20 nk_fraction_r lora 0.5644; pairs 40 nk_fraction_r lora 0.6993; pairs all nk_fraction_r lora 0.7019", + "source": "data/fig23_healthy.csv" + } + ], + "original_slide": 31, + "figures_before_palette": [ + "figures/fig3_healthy_pairs.png" + ] + }, + { + "kind": "content", + "section": "2", + "section_label": "Fig. 2e", + "text_source": { + "file": "data/fig23_healthy_slide_text.json", + "index": 3 + }, + "title": "Additional pairs improve donor-level information", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/fig2_healthy_pairs.png" + ], + "keywords": [ + "5 pairs: coverage 0.84 (zero-shot 0.19)", + "Donor NK fraction r 0.20 \u2192 0.70 with more pairs", + "5 pairs stop early, near 3,000 steps" + ], + "footnote": "Historical n + 7 validation design. The fixed-3000-step strict curve is a separate protocol update.", + "sources": "tools/fig23_healthy.py; data/fig23_healthy.csv", + "refs": [ + "Kynk\u00e4\u00e4nniemi 2019, NeurIPS", + "Naeem 2020, ICML" + ], + "refs_full": [ + "Kynk\u00e4\u00e4nniemi et al. Improved precision and recall metric for assessing generative models. NeurIPS 2019", + "Naeem et al. Reliable fidelity and diversity metrics for generative models. ICML 2020" + ], + "say": "Five pairs already give most of the coverage; more pairs mainly improve donor-level composition, such as the NK fraction.", + "footnote_full": "Historical n + 7 validation design. The fixed-3000-step strict curve is a separate protocol update.\nOriginal context: ImMicro, donor-level 5-fold CV, pooled scoring (107 donors); x = LoRA training pairs per fold (the last point is every training donor of the fold, 78\u201379). Calibration fitted on exactly the pairs the LoRA point saw. Selected step = early-stopping step on 7 inner-validation pairs (cap 20,000). Values: coverage LoRA 0.84 / 0.79 / 0.76 / 0.78 / 0.78; NK fraction r LoRA 0.20 / 0.40 / 0.56 / 0.70 / 0.70 at 5 / 10 / 20 / 40 / all pairs. Source: data/sec08_pairs_realism.csv, data/sec08_pairs_signal.csv, data/sec08_selected_steps_per_fold.csv.", + "numbers": [ + { + "value": "pairs 0 comp_l1 zero_shot 0.4549; pairs 5 comp_l1 zs_cal_budget 0.1916; pairs 10 comp_l1 zs_cal_budget 0.1996; pairs 20 comp_l1 zs_cal_budget 0.1911; pairs 40 comp_l1 zs_cal_budget 0.196; pairs all comp_l1 zs_cal_budget 0.1985; pairs 5 comp_l1 lora 0.1715; pairs 10 comp_l1 lora 0.1879; pairs 20 comp_l1 lora 0.1723; pairs 40 comp_l1 lora 0.1714; pairs all comp_l1 lora 0.1617; pairs 0 coverage zero_shot 0.1918; pairs 5 coverage zs_cal_budget 0.527; pairs 10 coverage zs_cal_budget 0.5287; pairs 20 coverage zs_cal_budget 0.5287; pairs 40 coverage zs_cal_budget 0.5224; pairs all coverage zs_cal_budget 0.5272; pairs 5 coverage lora 0.8392; pairs 10 coverage lora 0.7857; pairs 20 coverage lora 0.7552; pairs 40 coverage lora 0.7784; pairs all coverage lora 0.7843; pairs 0 nk_fraction_r zero_shot 0.4939; pairs 5 nk_fraction_r zs_cal_budget 0.5019; pairs 10 nk_fraction_r zs_cal_budget 0.515; pairs 20 nk_fraction_r zs_cal_budget 0.5251; pairs 40 nk_fraction_r zs_cal_budget 0.5; pairs all nk_fraction_r zs_cal_budget 0.5004; pairs 5 nk_fraction_r lora 0.1993; pairs 10 nk_fraction_r lora 0.3972; pairs 20 nk_fraction_r lora 0.5644; pairs 40 nk_fraction_r lora 0.6993; pairs all nk_fraction_r lora 0.7019; pairs 5 selected_step_median_k lora 3.25; pairs 10 selected_step_median_k lora 5.75; pairs 20 selected_step_median_k lora 19.5; pairs 40 selected_step_median_k lora 20; pairs all selected_step_median_k lora 20", + "source": "data/fig23_healthy.csv" + } + ], + "original_slide": 26, + "figures_before_palette": [ + "figures/fig2_healthy_pairs.png" + ] + }, + { + "kind": "content", + "section": "2", + "section_label": "Fig. 3a", + "title": "Measured bulk differs from pseudobulk", + "layout": "figure_keywords", + "figures": [ + "figures/sec05_condition_space.png" + ], + "keywords": [ + "Real bulk: 2.46 D from own pseudobulk", + "95% of squared offset is one shift", + "Shift removed (diagnostic): own donor 7.5% \u2192 75.7%" + ], + "footnote": "Existing paired-input diagnostic; mean-shift removal uses paired pseudobulks and is not a deployment method.", + "say": "The two pseudobulk clusters on PC1 are sc pools DB1-3 vs DB4-7, so the pairing lines cross mainly because of sc batch. Shift removal subtracts the mean offset of the other 106 donors; a bulk-only deployment would not have them.", + "sources": "tools/sec05_condition_space.py; data/sec05_condition_space_numbers.csv, data/sec05_condition_space_coords.csv", + "footnote_full": "Existing paired-input diagnostic; mean-shift removal uses paired pseudobulks and is not a deployment method.\nOriginal context: ImMicro, 107 healthy donors. Model inputs (log1p CPM, 5,508 genes), not learned ctx. D = median donor-to-donor distance; PCA-30 values are lower bounds (full space 3.19 D, 0.91 shared). Shift removal is a diagnostic that uses paired pseudobulks. Pseudobulk PC1 is sc pool batch (pool R\u00b2 0.96 on PC1).", + "original_slide": 17, + "figures_before_palette": [ + "figures/sec05_condition_space.png" + ] + }, + { + "kind": "content", + "section": "3", + "section_label": "Fig. 3b", + "text_source": { + "file": "data/fig23_healthy_slide_text.json", + "index": 5 + }, + "title": "Healthy donors: exactly five paired samples", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/fig3_healthy_fewpair.png" + ], + "keywords": [ + "LoRA-5 ahead on 6 of 6 metrics", + "Coverage 0.79 vs 0.04", + "E_N 1.5 vs 3.6 (scSemiProfiler's own metric)" + ], + "footnote": "ImMicro: exactly 5 pairs, no extra validation, 3000-step LoRA; five draws, remaining 102 donors per draw.", + "sources": "tools/fig23_healthy.py; data/fig23_healthy.csv", + "refs": [ + "Wang 2024, Nat Commun", + "Lopez-Paz 2017, ICLR" + ], + "refs_full": [ + "Wang et al. scSemiProfiler: advancing large-scale single-cell studies through semi-profiling with deep generative models and active learning. Nat Commun 15:5989 (2024). doi:10.1038/s41467-024-50150-1", + "Lopez-Paz & Oquab. Revisiting classifier two-sample tests. ICLR 2017" + ], + "say": "With the same five healthy donors, LoRA-5 beats scSemiProfiler on all six cell-level metrics, including scSemiProfiler's own error metric.", + "footnote_full": "ImMicro: exactly 5 pairs, no extra validation, 3000-step LoRA; five draws, remaining 102 donors per draw.\nOriginal context: Main comparison on healthy donors: per draw, LoRA-5 and scSemiProfiler get the same 5 paired donors (scSemiProfiler: as its representatives) and are tested on the other 102 from their real bulk; 5 draws. Zero-shot (no pairs) shown for reference. Dashed: real vs real cells. Values (LoRA-5 / scSemiProfiler-5): coverage 0.792 / 0.036; precision 0.952 / 0.518; comp_l1 0.176 / 0.233; c2st_knn_auc 0.943 / 0.999; mmd2_bw1 0.023 / 0.186; en_all_sq 1.547 / 3.649. Source: data/sec09_fewpair_cells.csv, data/sec09_extra_fewpair.csv.", + "numbers": [ + { + "value": "5 donors (mean of 5 draws) coverage zs_raw 0.19; 5 donors (mean of 5 draws) coverage scsemi 0.03583; 5 donors (mean of 5 draws) coverage lora 0.7925; 5 donors (mean of 5 draws) precision zs_raw 0.6772; 5 donors (mean of 5 draws) precision scsemi 0.518; 5 donors (mean of 5 draws) precision lora 0.9521; 5 donors (mean of 5 draws) comp_l1 zs_raw 0.4451; 5 donors (mean of 5 draws) comp_l1 scsemi 0.2325; 5 donors (mean of 5 draws) comp_l1 lora 0.1756; 5 donors (mean of 5 draws) c2st_knn_auc zs_raw 0.9966; 5 donors (mean of 5 draws) c2st_knn_auc scsemi 0.9988; 5 donors (mean of 5 draws) c2st_knn_auc lora 0.9427; 5 donors (mean of 5 draws) mmd2_bw1 zs_raw 0.1026; 5 donors (mean of 5 draws) mmd2_bw1 scsemi 0.1865; 5 donors (mean of 5 draws) mmd2_bw1 lora 0.02308; 5 donors (mean of 5 draws) en_all_sq zs_raw 7.236; 5 donors (mean of 5 draws) en_all_sq scsemi 3.649; 5 donors (mean of 5 draws) en_all_sq lora 1.547", + "source": "data/fig23_healthy.csv" + } + ], + "original_slide": 30, + "figures_before_palette": [ + "figures/fig3_healthy_fewpair.png" + ] + }, + { + "kind": "content", + "title": "Composition: error and donor-level wins", + "section_label": "Fig. 3c", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/review_style_20261013_healthy_composition.png" + ], + "keywords": [], + "footnote": "Exactly five pairs per draw. Absolute errors averaged within each held-out donor; 107 donors, equal weight. Descriptive.", + "footnote_full": "Exactly five pairs per draw. Absolute errors averaged within each held-out donor; 107 donors, equal weight. Descriptive.", + "sources": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_healthy_composition.pdf", + "figures_before_palette": [ + "figures/review_style_20261013_healthy_composition.png" + ] + }, + { + "kind": "content", + "section": "3", + "section_label": "Fig. 3d", + "text_source": { + "file": "data/fig23_healthy_slide_text.json", + "index": 7 + }, + "title": "Representative-dependent composition", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/fig3_healthy_copy.png" + ], + "keywords": [ + "Composition varies with the selected representative", + "Compare generated fractions with held-out observed fractions" + ], + "footnote": "Existing healthy five-draw diagnostic. This evaluates composition dependence, not literal copying of every cell.", + "sources": "tools/fig23_healthy.py; data/fig23_healthy.csv", + "refs": [ + "Wang 2024, Nat Commun" + ], + "refs_full": [ + "Wang et al. scSemiProfiler: advancing large-scale single-cell studies through semi-profiling with deep generative models and active learning. Nat Commun 15:5989 (2024). doi:10.1038/s41467-024-50150-1" + ], + "say": "scSemiProfiler does worse than simply copying its representative's composition, and its error follows that representative; LoRA does better than the copy.", + "footnote_full": "Existing healthy five-draw diagnostic. This evaluates composition dependence, not literal copying of every cell.\nOriginal context: ImMicro, 5 donors. scSemiProfiler assigns each test sample one representative; 'copy' uses that representative's observed composition as the prediction. Right: Spearman over test samples between each method's composition error and the copy's error. Source: data/sec09_scsemi_representative.csv (scsemi/analysis/representative_dependence_20261002).", + "numbers": [ + { + "value": "5 donors (mean of 5 draws) comp_l1 copy 0.1871; 5 donors (mean of 5 draws) comp_l1 scsemi 0.2325; 5 donors (mean of 5 draws) comp_l1 lora 0.1756; spearman_with_copy_error scsemi 0.5736; spearman_with_copy_error lora 0.3509", + "source": "data/fig23_healthy.csv" + } + ], + "original_slide": 32, + "figures_before_palette": [ + "figures/fig3_healthy_copy.png" + ] + }, + { + "kind": "content", + "section": "3", + "section_label": "Fig. 3e", + "text_source": { + "file": "data/sec09_time_to_apply_slide_text.json", + "index": 0 + }, + "title": "Adaptation and generation time", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/sec09_time_to_apply.png" + ], + "keywords": [ + "LoRA-5: 3\u201311 min per draw, every test sample", + "scSemiProfiler-5: 57 min\u20134.1 h per draw", + "17\u201359\u00d7 less GPU time, same 5 donors" + ], + "footnote": "Existing measurements: adaptation + target generation; pretraining, GPU wait and scoring excluded.", + "sources": "tools/sec09_time_to_apply.py; data/sec09_time_to_apply.csv, data/sec09_time_to_apply_runs.csv; data/sec06_ft_method.csv (pretrain_wall_hours)", + "footnote_full": "Existing measurements: adaptation + target generation; pretraining, GPU wait and scoring excluded.\nOriginal context: Same 25 draws as the '5 donors' slides (5 cohorts x 5 draws; same 5 training donors and test samples for both methods). GPU time on one GB10, every step alone on the GPU (shared lock). LoRA-5: fine-tune = RUN.json wall_seconds of the 3,000-step LoRA; generate = wall clock of the sampling job in the queue log (model load + DDIM 50, 1,500 cells per test sample). scSemiProfiler-5 (upstream c2aea02, default budget): stage 1 + stage 2 pretraining of every representative (model receipts) + its per-target inference (5 ministages x 150 epochs, export receipt total_target_seconds). Not counted on either side: CPU data preparation, scoring, work lost to interruptions, and PBISC's one-time pretraining (15.9 h; 'Fine-tuning with paired bulk + sc' slide), which every cohort reuses. scSemiProfiler time grows with the number of test samples because it trains one model per sample. Per-draw ratio scSemiProfiler / LoRA: ImMicro 15\u201319\u00d7; Sepsis 26\u201347\u00d7; Imperial 19\u201326\u00d7; Giroux 54\u201363\u00d7; Ji 11\u201320\u00d7 (all 25 draws 11\u201363\u00d7; keyword and figure = mean of each cohort's 5 draws).", + "numbers": [ + { + "value": "ImMicro: LoRA-5 8.4 min (fine-tune 1.9 + generate 6.5), scSemiProfiler-5 141 min (representatives 15 + per-sample models 126); 15\u201319\u00d7 per draw; per test sample 4.9 s vs 83 s", + "source": "data/sec09_time_to_apply.csv" + }, + { + "value": "Sepsis: LoRA-5 6.7 min (fine-tune 1.8 + generate 4.9), scSemiProfiler-5 244 min (representatives 30 + per-sample models 214); 26\u201347\u00d7 per draw; per test sample 5.2 s vs 190 s", + "source": "data/sec09_time_to_apply.csv" + }, + { + "value": "Imperial: LoRA-5 10.8 min (fine-tune 1.8 + generate 9.0), scSemiProfiler-5 229 min (representatives 32 + per-sample models 197); 19\u201326\u00d7 per draw; per test sample 4.5 s vs 96 s", + "source": "data/sec09_time_to_apply.csv" + }, + { + "value": "Giroux: LoRA-5 2.6 min (fine-tune 1.9 + generate 0.7), scSemiProfiler-5 156 min (representatives 117 + per-sample models 38); 54\u201363\u00d7 per draw; per test sample 14.8 s vs 872 s", + "source": "data/sec09_time_to_apply.csv" + }, + { + "value": "Ji: LoRA-5 3.4 min (fine-tune 1.8 + generate 1.6), scSemiProfiler-5 57 min (representatives 24 + per-sample models 33); 11\u201320\u00d7 per draw; per test sample 8.1 s vs 137 s", + "source": "data/sec09_time_to_apply.csv" + }, + { + "value": "PBISC one-time pretraining 15.9 h (not counted per draw)", + "source": "data/sec06_ft_method.csv (pretrain_wall_hours)" + } + ], + "refs": [ + "Wang 2024, Nat Commun" + ], + "refs_full": [ + "Wang et al. scSemiProfiler: advancing large-scale single-cell studies through semi-profiling with deep generative models and active learning. Nat Commun 15:5989 (2024). doi:10.1038/s41467-024-50150-1" + ], + "original_slide": 34, + "figures_before_palette": [ + "figures/sec09_time_to_apply.png" + ] + }, + { + "kind": "content", + "section": "4", + "section_label": "Fig. 4a", + "sec09_index": 0, + "title": "Paired cohorts and evaluation units", + "layout": "full", + "figures": [ + "figures/sec09_cohorts.png" + ], + "keywords": [ + "403 paired samples, 275 donors", + "Identity screening where available; donor-level splits", + "Ji bulk: whole blood, not PBMC" + ], + "footnote": "Existing paired-cohort map. Five donors means 5 pairs in Sepsis, 12-15 Imperial, 12-14 Giroux, 7-9 Ji.", + "footnote_full": "Existing paired-cohort map. Five donors means 5 pairs in Sepsis, 12-15 Imperial, 12-14 Giroux, 7-9 Ji.\nOriginal context: Usable pair = passed the pairing identity check and >= 200 cells (fold >= 0 in data//folds.csv); dropped 15 identity-fail, 11 borderline and 1 low-cell pairs of 430. ImMicro: no identity flag, all used. Imperial, Giroux, Ji: several pairs per donor. Case vs control = samples in scores//pooled/summary.json. Giroux 'moderate' = GEO 'severe' (paper term); its moderate vs mild subjects are 4F/0M vs 0F/5M (scored), so its disease metrics exclude sex-linked genes. Labels: data//QC_REPORT.md; ImMicro: GEO GSE314922/GSE314416.", + "sources": "tools/sec09_paired_validation.py; data/sec09_cohorts.csv, data/sec09_quoted_numbers.csv, data/sec09_slide_text.json", + "numbers": [ + { + "value": "403 paired samples (sum of usable rows)", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data//fewshot_n5/DRAWS.json (n_usable_rows)" + }, + { + "value": "275 donors (sum of usable donors)", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data//fewshot_n5/DRAWS.json (n_usable_donors)" + }, + { + "value": "dropped 15 identity-fail, 11 borderline, 1 < 200 cells, of 430", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data//features.npz (identity_pass, identity_check); samples.csv (n_cells); /home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/tools/prepare_dataset.py (MIN_CELLS)" + }, + { + "value": "ImMicro 107 pairs / 107 donors", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data/immicro/fewshot_n5/DRAWS.json" + }, + { + "value": "Sepsis 82 pairs / 82 donors; sepsis 61 vs control 21", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data/ye2026_sepsis/fewshot_n5/DRAWS.json; /home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores/ye2026_sepsis/pooled/summary.json" + }, + { + "value": "Imperial 157 pairs / 57 donors; COVID+ 113 vs COVID\u2212 34 (+10 recovery, neither)", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data/imperial_eskd/fewshot_n5/DRAWS.json; /home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores/imperial_eskd/pooled/summary.json" + }, + { + "value": "Giroux 24 pairs / 9 donors; moderate 11 vs mild 13", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data/giroux2022/fewshot_n5/DRAWS.json; /home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores/giroux2022/pooled/summary.json" + }, + { + "value": "Ji 33 pairs / 20 donors; ChILI 18 vs ICI control 15", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data/ji2025_chili/fewshot_n5/DRAWS.json; /home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores/ji2025_chili/pooled/summary.json" + } + ], + "original_slide": 37, + "figures_before_palette": [ + "figures/sec09_cohorts.png" + ] + }, + { + "kind": "content", + "section": "4", + "section_label": "Fig. 4b", + "text_source": { + "file": "data/fig4_disease_slide_text.json", + "index": 0 + }, + "title": "Disease signals: five-donor draws", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/fig4_disease_fewpair.png" + ], + "keywords": [ + "vs scSemiProfiler-5: logFC r 13/20, partial r 14/20 draws", + "Beyond bulk: partial r 0.60 vs 0.43 (Imperial)", + "Above zero-shot and calibration in 4/4 cohorts" + ], + "footnote": "Existing five-donor protocol; sample counts differ by cohort. Original sample-level estimator; corrected update separate.", + "sources": "tools/fig4_disease.py; data/fig4_disease.csv", + "refs": [ + "Wang 2024, Nat Commun", + "Vi\u00f1as Torn\u00e9 2026, Nat Biotechnol" + ], + "refs_full": [ + "Wang et al. scSemiProfiler: advancing large-scale single-cell studies through semi-profiling with deep generative models and active learning. Nat Commun 15:5989 (2024). doi:10.1038/s41467-024-50150-1", + "Vi\u00f1as Torn\u00e9 et al. Systema: a framework for evaluating genetic perturbation response prediction beyond systematic variation. Nat Biotechnol 44:1050-1059 (2026). doi:10.1038/s41587-025-02777-8" + ], + "say": "The main comparison uses five pairs, the situation a lab is in. With the same five donors, LoRA-5 is above both zero-shot versions in every cohort and ahead of scSemiProfiler in most draws. The bold panel is partial r: the cell-type signal the bulk itself does not carry, which is what single cells add.", + "footnote_full": "Existing five-donor protocol; sample counts differ by cohort. Original sample-level estimator; corrected update separate.\nOriginal context: Main comparison: 5 donors per draw, LoRA\u20135 vs scSemiProfiler with the same 5 donors as representatives (and the two zero\u2013shot PBISC versions), 5 draws per cohort, every other usable sample tested. logFC r = Pearson r of case\u2013vs\u2013control logFC (mean CP10K) with the observed cells, mean over T, B, monocyte and NK. Partial r = the same after both logFCs are residualised on the input bulk's logFC: the cell\u2013type signal the bulk does not carry. Top\u201350 DEG: LoRA\u20135 ahead in 12/20 draws, tied in 8 (Giroux and Ji: few DEG in the observed cells, mostly 0 for every method). Giroux's 5\u2013donor test sets hold 10\u201312 samples from 4 subjects (2 vs 2), so its disease metrics are noisy; there scSemiProfiler\u20135 is ahead. Giroux: sex\u2013linked genes excluded (severity is confounded with sex). Per cohort, mean over draws (r / partial r / DEG): Sepsis LoRA\u20135 0.35 / 0.30 / 16.4, scSemiProfiler\u20135 0.40 / 0.33 / 5.2; Imperial LoRA\u20135 0.75 / 0.60 / 113.6, scSemiProfiler\u20135 0.56 / 0.43 / 46.8; Giroux LoRA\u20135 0.30 / 0.19 / 0.2, scSemiProfiler\u20135 0.40 / 0.24 / 0.0; Ji LoRA\u20135 0.33 / 0.30 / 0.2, scSemiProfiler\u20135 0.15 / 0.09 / 0.0. Source: data/fig4_disease.csv (scores[_nosex]//fewshot_n5/draw[_scsemi5]).", + "numbers": [ + { + "value": "Sepsis LoRA\u20135 0.35 / 0.30 / 16.4, scSemiProfiler\u20135 0.40 / 0.33 / 5.2; Imperial LoRA\u20135 0.75 / 0.60 / 113.6, scSemiProfiler\u20135 0.56 / 0.43 / 46.8; Giroux LoRA\u20135 0.30 / 0.19 / 0.2, scSemiProfiler\u20135 0.40 / 0.24 / 0.0; Ji LoRA\u20135 0.33 / 0.30 / 0.2, scSemiProfiler\u20135 0.15 / 0.09 / 0.0", + "source": "data/fig4_disease.csv" + } + ], + "original_slide": 38, + "figures_before_palette": [ + "figures/fig4_disease_fewpair.png" + ] + }, + { + "kind": "content", + "section": "4", + "section_label": "Fig. 4c", + "text_source": { + "file": "data/fig4_disease_slide_text.json", + "index": 1 + }, + "title": "Disease signals: five-fold validation", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/fig4_disease_fivefold.png" + ], + "keywords": [ + "logFC r up to 0.82 (Imperial)", + "vs scSemiProfiler: r 3/4, partial r 3/4, DEG 4/4 cohorts", + "Top PBISC version on logFC r in 4/4 cohorts" + ], + "footnote": "Original donor-level split and original signal estimator. These results are distinct from the exact-five update.", + "sources": "tools/fig4_disease.py; data/fig4_disease.csv", + "refs": [ + "Wang 2024, Nat Commun", + "Vi\u00f1as Torn\u00e9 2026, Nat Biotechnol" + ], + "refs_full": [ + "Wang et al. scSemiProfiler: advancing large-scale single-cell studies through semi-profiling with deep generative models and active learning. Nat Commun 15:5989 (2024). doi:10.1038/s41467-024-50150-1", + "Vi\u00f1as Torn\u00e9 et al. Systema: a framework for evaluating genetic perturbation response prediction beyond systematic variation. Nat Biotechnol 44:1050-1059 (2026). doi:10.1038/s41587-025-02777-8" + ], + "say": "With every fold's pairs, LoRA reaches a disease logFC r of 0.82 in Imperial and is the best PBISC version on logFC r in every cohort. This is reported separately from the five-pair comparison.", + "footnote_full": "Original donor-level split and original signal estimator. These results are distinct from the exact-five update.\nOriginal context: Reported separately from the 5-pair comparison: donor-level 5-fold CV, each sample generated by the fold model that never trained on it, one pooled scoring call per method; scSemiProfiler uses exactly LoRA's training pairs as representatives. Per cohort LoRA vs scSemiProfiler (r / partial r / DEG): Sepsis 0.53 / 0.47 / 23 vs 0.44 / 0.36 / 6; Imperial 0.82 / 0.69 / 143 vs 0.71 / 0.54 / 87; Giroux 0.46 / 0.32 / 5 vs 0.54 / 0.39 / 0; Ji 0.48 / 0.45 / 15 vs 0.27 / 0.23 / 0. Giroux: sex-linked genes excluded. Source: data/fig4_disease.csv (scores[_nosex]//pooled, scsemi_cv_20261002/pooled; sepsis DEG from pooled_deg).", + "numbers": [ + { + "value": "Sepsis 0.53 / 0.47 / 23 vs 0.44 / 0.36 / 6; Imperial 0.82 / 0.69 / 143 vs 0.71 / 0.54 / 87; Giroux 0.46 / 0.32 / 5 vs 0.54 / 0.39 / 0; Ji 0.48 / 0.45 / 15 vs 0.27 / 0.23 / 0", + "source": "data/fig4_disease.csv" + } + ], + "original_slide": 39, + "figures_before_palette": [ + "figures/fig4_disease_fivefold.png" + ] + }, + { + "kind": "content", + "section": "4", + "section_label": "Fig. 4d", + "text_source": { + "file": "data/sec09_cat_curves_slide_text.json", + "index": 0 + }, + "title": "Disease-gene agreement across list lengths", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/sec09_cat_curves.png" + ], + "keywords": [ + "LoRA top model: 3/4 (5 donors), 2/4 (5-fold)", + "LoRA-5 above scSemiProfiler-5: 3/4 cohorts (k = 100)", + "Chance at k = 100: 0.012\u20130.016" + ], + "footnote": "Existing CAT curves; ranking agreement and chance line. Patient-aware reanalysis reuses the same predictions.", + "sources": "tools/sec09_cat_curves.py; data/sec09_cat_curves.csv", + "refs": [ + "Irizarry 2005, Nat Methods", + "Webber 2010, ACM TOIS", + "Squair 2021, Nat Commun" + ], + "refs_full": [ + "Irizarry et al. Multiple-laboratory comparison of microarray platforms. Nat Methods 2:345-350 (2005). doi:10.1038/nmeth756", + "Webber, Moffat & Zobel. A similarity measure for indefinite rankings. ACM Trans Inf Syst 28(4):20 (2010). doi:10.1145/1852102.1852106", + "Squair et al. Confronting false discoveries in single-cell differential expression. Nat Commun 12:5692 (2021). doi:10.1038/s41467-021-25960-2" + ], + "footnote_full": "Existing CAT curves; ranking agreement and chance line. Patient-aware reanalysis reuses the same predictions.\nOriginal context: Correspondence at the top (Irizarry et al. 2005): per cell type, every gene tested on both sides is ranked by |limma-voom moderated t| (case vs control, same donor x type pseudobulks as the DEG panels); CAT(k) = share of the observed top-k genes that are also in the generated top-k with the same logFC sign. Mean over T / B / Mono / NK; 5 donors = mean over 5 draws (scSemiProfiler: its 25 draws), the main comparison; 5-fold = pooled call, shown separately. Chance = k / (2G), G = genes tested on both sides (0.012\u20130.016 at k = 100). No significance threshold, so cohorts where few genes pass FDR (Giroux, Ji) still get a curve. Rank-biased overlap (Webber et al. 2010, p = 0.98) gives the same order: scores_contrast/{fewshot,pooled}_rbo.csv. Giroux: sex-linked genes excluded. Sepsis CAT100 5 donors LoRA 0.092 / scSemi 0.037 / calibration 0.069 / raw 0.028 / chance 0.012; 5-fold LoRA 0.163; Imperial CAT100 5 donors LoRA 0.243 / scSemi 0.133 / calibration 0.198 / raw 0.179 / chance 0.012; 5-fold LoRA 0.270; Giroux CAT100 5 donors LoRA 0.048 / scSemi 0.075 / calibration 0.050 / raw 0.048 / chance 0.016; 5-fold LoRA 0.068; Ji CAT100 5 donors LoRA 0.091 / scSemi 0.030 / calibration 0.021 / raw 0.021 / chance 0.014; 5-fold LoRA 0.152.", + "numbers": [ + { + "value": "Sepsis CAT100 5 donors LoRA 0.092 / scSemi 0.037 / calibration 0.069 / raw 0.028 / chance 0.012; 5-fold LoRA 0.163; Imperial CAT100 5 donors LoRA 0.243 / scSemi 0.133 / calibration 0.198 / raw 0.179 / chance 0.012; 5-fold LoRA 0.270; Giroux CAT100 5 donors LoRA 0.048 / scSemi 0.075 / calibration 0.050 / raw 0.048 / chance 0.016; 5-fold LoRA 0.068; Ji CAT100 5 donors LoRA 0.091 / scSemi 0.030 / calibration 0.021 / raw 0.021 / chance 0.014; 5-fold LoRA 0.152", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores_contrast/fewshot_cat.csv; /home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores_contrast/pooled_cat.csv" + } + ], + "original_slide": 40, + "figures_before_palette": [ + "figures/sec09_cat_curves.png" + ] + }, + { + "kind": "content", + "section": "4", + "section_label": "Fig. 4e", + "text_source": { + "file": "data/secS_deconvolution_slide_text.json", + "index": 1 + }, + "title": "Cell-type expression: deconvolution comparison", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/secS_deconv_disease.png" + ], + "keywords": [ + "logFC r: PBISC ahead in 15/20 draws (BayesPrism), 17/17 (CIBERSORTx)", + "Beyond bulk: PBISC ahead in 18/20 and 17/17", + "Top-50 DEG mixed: PBISC ahead in 6/20 and 6/17" + ], + "footnote": "Completed BayesPrism/CIBERSORTx comparisons; same five-donor support, method-supported expression metrics.", + "sources": "tools/secS_deconvolution.py; data/secS_deconvolution.csv; plans/bayesprism_paired_20261006/results/{RESULTS_full.md, RESULTS_cibersortx.md, TABLE_*_full.csv, TABLE_*_cibersortx.csv}", + "refs": [ + "Chu 2022, Nat Cancer", + "Newman 2019, Nat Biotechnol" + ], + "refs_full": [ + "Chu et al. Cell type and gene expression deconvolution with BayesPrism enables Bayesian integrative analysis across bulk and single-cell RNA sequencing in oncology. Nat Cancer 3:505-517 (2022). doi:10.1038/s43018-022-00356-3", + "Newman et al. Determining cell type abundance and expression from bulk tissues with digital cytometry. Nat Biotechnol 37:773-782 (2019). doi:10.1038/s41587-019-0114-2" + ], + "say": "Both methods also estimate expression per cell type: BayesPrism's posterior and CIBERSORTx HiRes. Their case-versus-control changes per cell type agree less with the truth than PBISC's in most draws (15 of 20 for BayesPrism, all 17 scored for CIBERSORTx; BayesPrism is ahead in Giroux). Beyond what the bulk itself carries, they add little: partial r at most 0.19, against 0.19 to 0.60 for PBISC. Top-50 DEG overlap is mixed: deconvolution is ahead in sepsis, PBISC in Imperial, and Ji and Giroux are near zero for every method.", + "footnote_full": "Completed BayesPrism/CIBERSORTx comparisons; same five-donor support, method-supported expression metrics.\nOriginal context: Cell-type deconvolution, not single cells. Case-vs-control per-cell-type logFC (mean CP10K; T, B, Monocyte, NK) from BayesPrism's posterior cell-type expression Z and CIBERSORTx HiRes, scored with the benchmark's own functions; partial r = the same r after both logFCs are residualised on the input bulk's logFC (the cell-type signal the bulk does not carry); top-50 DEG = Welch, summed over types (many ties at 0 in Ji and Giroux). 5-donor draws (the 5-pair comparison), mean over draws with a score. Giroux excludes sex-linked genes, as everywhere in the deck. CIBERSORTx: official Docker images (x86, run under qemu emulation on server_1), single-cell reference mode with S-mode batch correction (up to 400 reference cells per type), fractions in cell units (divided by each type's mean UMI per reference cell). CIBERSORTx HiRes gives per-sample cell-type expression for the genes it imputes per type (others are written as 1 and count as zero here); it imputed no gene of the four contrast types in 1 Ji ChILI and 2 Giroux draws, which therefore have no CIBERSORTx disease score. Source: plans/bayesprism_paired_20261006/results/RESULTS_full.md, RESULTS_cibersortx.md.", + "numbers": [ + { + "value": "Sepsis: logFC r PBISC 0.350 / BayesPrism 0.241 / CIBERSORTx 0.069; partial r 0.305 / 0.095 / -0.010; DEG 16.4 / 27.0 / 25.4; Ji ChILI: logFC r PBISC 0.333 / BayesPrism 0.193 / CIBERSORTx 0.158; partial r 0.300 / 0.032 / 0.005; DEG 0.2 / 2.0 / 0.0; Imperial: logFC r PBISC 0.749 / BayesPrism 0.598 / CIBERSORTx 0.283; partial r 0.603 / 0.187 / 0.061; DEG 113.6 / 72.4 / 30.8; Giroux: logFC r PBISC 0.297 / BayesPrism 0.371 / CIBERSORTx 0.095; partial r 0.189 / 0.117 / 0.029; DEG 0.2 / 0.2 / 0.0", + "source": "data/secS_deconvolution.csv" + }, + { + "value": "draws where PBISC LoRA-5 beats BayesPrism: logFC r 15/20, partial r 18/20, top-50 DEG 6 (BayesPrism ahead 8, ties 6); beats CIBERSORTx: logFC r 17/17, partial r 17/17, top-50 DEG 6 (CIBERSORTx ahead 4, ties 7)", + "source": "plans/bayesprism_paired_20261006/results/TABLE_fewshot_full.csv, TABLE_fewshot_cibersortx.csv" + } + ], + "original_slide": 41, + "figures_before_palette": [ + "figures/secS_deconv_disease.png" + ] + }, + { + "kind": "content", + "section": "4", + "section_label": "Fig. 4f", + "text_source": { + "file": "data/secS_deconvolution_slide_text.json", + "index": 0 + }, + "title": "Composition: deconvolution comparison", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/secS_deconv_composition.png" + ], + "keywords": [ + "PBISC closer than BayesPrism in 22/25 draws, than CIBERSORTx in 24/25", + "Composition compared with held-out observed cells", + "Five-fold: PBISC lowest error in 4 of 5 cohorts" + ], + "footnote": "Completed deconvolution outputs already exist. Verified identical Sepsis supports use adoption instead of method reruns.", + "sources": "tools/secS_deconvolution.py; data/secS_deconvolution.csv; plans/bayesprism_paired_20261006/results/{RESULTS_full.md, RESULTS_cibersortx.md, TABLE_*_full.csv, TABLE_*_cibersortx.csv}", + "refs": [ + "Chu 2022, Nat Cancer", + "Newman 2019, Nat Biotechnol", + "Wang 2024, Nat Commun" + ], + "refs_full": [ + "Chu et al. Cell type and gene expression deconvolution with BayesPrism enables Bayesian integrative analysis across bulk and single-cell RNA sequencing in oncology. Nat Cancer 3:505-517 (2022). doi:10.1038/s43018-022-00356-3", + "Newman et al. Determining cell type abundance and expression from bulk tissues with digital cytometry. Nat Biotechnol 37:773-782 (2019). doi:10.1038/s41587-019-0114-2", + "Wang et al. scSemiProfiler: advancing large-scale single-cell studies through semi-profiling with deep generative models and active learning. Nat Commun 15:5989 (2024). doi:10.1038/s41467-024-50150-1" + ], + "say": "For reference only. Could standard bulk deconvolution do the same job with the same paired donors? BayesPrism and CIBERSORTx used the paired donors' own cells as the reference and the test samples' real bulk as the mixture, scored with the same truth and code as PBISC. PBISC estimates the cell-type mix more closely in almost every draw. On pseudobulk BayesPrism recovers the right mix, so the gap comes from real bulk, which PBISC learns from the pairs. Deconvolution also returns proportions, not cells.", + "footnote_full": "Completed deconvolution outputs already exist. Verified identical Sepsis supports use adoption instead of method reruns.\nOriginal context: Cell-type deconvolution (cell-type fractions and per-type expression, not single cells) given the same paired donors as PBISC LoRA, on the paired benchmark's own splits, observed truth and scorer: 5-donor draws (reference = the draw's 5 training donors' observed cells, mixture = the test samples' real bulk) and 5-fold CV (reference = each fold's LoRA training + inner-validation pairs; pooled). BayesPrism: recommended gene set (protein-coding minus chrX/Y/M and ribosomal genes, 15-19k), reference states = gate | AIFI L2; its better theta (initial or final) in cell units is shown. CIBERSORTx: official Docker images (x86, run under qemu emulation on server_1), single-cell reference mode with S-mode batch correction (up to 400 reference cells per type), fractions in cell units (divided by each type's mean UMI per reference cell). CIBERSORTx HiRes gives per-sample cell-type expression for the genes it imputes per type (others are written as 1 and count as zero here); it imputed no gene of the four contrast types in 1 Ji ChILI and 2 Giroux draws, which therefore have no CIBERSORTx disease score. Composition L1 over the six-type gate, lower is better. 5-fold: PBISC has the lowest L1 except in Giroux (scSemiProfiler). Pipeline check: on pseudobulk mixtures BayesPrism recovers the observed mRNA shares (e.g. T 0.39 vs 0.38), so the error comes from real bulk. Source: plans/bayesprism_paired_20261006/results/RESULTS_full.md, RESULTS_cibersortx.md.", + "numbers": [ + { + "value": "ImMicro: 5 donors L1 PBISC 0.176 / BayesPrism 0.537 / CIBERSORTx 0.491; 5-fold PBISC 0.162 / BayesPrism 0.575 / CIBERSORTx 0.507 / scSemiProfiler 0.217; Sepsis: 5 donors L1 PBISC 0.466 / BayesPrism 0.446 / CIBERSORTx 0.582; 5-fold PBISC 0.362 / BayesPrism 0.475 / CIBERSORTx 0.607 / scSemiProfiler 0.502; Ji ChILI: 5 donors L1 PBISC 0.513 / BayesPrism 0.893 / CIBERSORTx 0.827; 5-fold PBISC 0.504 / BayesPrism 0.883 / CIBERSORTx 0.914 / scSemiProfiler 0.575; Imperial: 5 donors L1 PBISC 0.503 / BayesPrism 0.845 / CIBERSORTx 0.646; 5-fold PBISC 0.376 / BayesPrism 0.887 / CIBERSORTx 0.592 / scSemiProfiler 0.520; Giroux: 5 donors L1 PBISC 0.256 / BayesPrism 1.455 / CIBERSORTx 0.669; 5-fold PBISC 0.258 / BayesPrism 0.878 / CIBERSORTx 0.591 / scSemiProfiler 0.243", + "source": "data/secS_deconvolution.csv" + } + ], + "original_slide": 46, + "figures_before_palette": [ + "figures/secS_deconv_composition.png" + ] + }, + { + "kind": "content", + "title": "Rare-state frequencies in observed cells", + "section_label": "Fig. 5a", + "layout": "full", + "figures": [ + "figures/review_style_20261013_rare_observed.png" + ], + "keywords": [], + "footnote": "Original five-donor outputs. Median across draws; state share within its parent population. Controls included.", + "footnote_full": "Original five-donor outputs. Median across draws; state share within its parent population. Controls included.", + "sources": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_rare_observed.pdf", + "figures_before_palette": [ + "figures/review_style_20261013_rare_observed.png" + ] + }, + { + "kind": "content", + "title": "Rare-state fidelity: zero-aware reanalysis", + "section_label": "Fig. 5b", + "layout": "full", + "figures": [ + "figures/review_style_20261013_rare_corrected_comparison.png" + ], + "keywords": [], + "footnote": "Original five-donor outputs; zero draws retained and actual donor IDs. 59 state/cohort combinations, including controls.", + "footnote_full": "Original five-donor outputs; zero draws retained and actual donor IDs. 59 state/cohort combinations, including controls.", + "sources": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_rare_corrected_comparison.pdf", + "figures_before_palette": [ + "figures/review_style_20261013_rare_corrected_comparison.png" + ] + }, + { + "kind": "content", + "title": "Behcet: one-gene and multi-gene C1Q rules", + "section_label": "Fig. 5c", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/review_style_20261013_behcet.png" + ], + "keywords": [], + "footnote": "Descriptive: unmatched reference includes training donors. Same three pairs; two identity-pass and one uncertain.", + "footnote_full": "Descriptive: unmatched reference includes training donors. Same three pairs; two identity-pass and one uncertain.", + "sources": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_behcet.pdf", + "figures_before_palette": [ + "figures/review_style_20261013_behcet.png" + ] + }, + { + "kind": "status", + "title": "Remaining work", + "section_label": "Figure status", + "rows": [ + [ + "Protocol updates", + "Total n = 5/10/20/40 without extra validation; exact-five samples in multi-visit cohorts." + ], + [ + "Analysis of existing outputs", + "Patient-aware contrasts, rare-state scoring and healthy cell visualizations." + ], + [ + "Figure 6: new data", + "Reproduce a published finding in the user-prepared bulk-only cohort." + ] + ], + "footnote": "Existing results remain visible. Protocol updates and missing-data work are identified separately." + }, + { + "kind": "content", + "section": "2", + "section_label": "Supporting", + "title": "Control-cell embedding diagnostic", + "layout": "two_figures", + "figures": [ + "figures/sec05_cells_umap.png", + "figures/method_palette_20261013/sec05_cells_metrics.png" + ], + "keywords": [ + "Monocytes: raw 30.6% vs real 16.7%", + "Same noise: 19% of T cells become monocytes", + "Precision: raw 0.36 vs pseudobulk 0.92" + ], + "footnote": "Existing same-noise input diagnostic; not the new LoRA/scSemi OOF embedding. UMAP overlap is not a quantitative score.", + "footnote_full": "Existing same-noise input diagnostic; not the new LoRA/scSemi OOF embedding. UMAP overlap is not a quantitative score.\nOriginal context: ImMicro, 107 donors. Final model f01 6k 1\u00d7 E16 (DDIM 50, EMA); each cell starts from the same noise under every input. Precision/coverage from one scoring call. Composition L1: raw 0.455, pseudobulk 0.166, cohort-mean null 0.161. The UMAP is fitted on real cells and overstates overlap.", + "sources": "tools/sec05_cells_offmanifold.py; data/sec05_cells_metrics.csv, data/sec05_cells_umap_coords.csv", + "refs": [ + "Kynk\u00e4\u00e4nniemi 2019, NeurIPS", + "Naeem 2020, ICML" + ], + "refs_full": [ + "Kynk\u00e4\u00e4nniemi et al. Improved precision and recall metric for assessing generative models. NeurIPS 2019", + "Naeem et al. Reliable fidelity and diversity metrics for generative models. ICML 2020" + ], + "original_slide": 19, + "figures_before_palette": [ + "figures/sec05_cells_umap.png", + "figures/sec05_cells_metrics.png" + ] + }, + { + "kind": "content", + "section": "S", + "section_label": "Supporting", + "sec09_index": 1, + "title": "Generated-cell quality across cohorts", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/sec09_fewpair_cells.png" + ], + "keywords": [ + "LoRA-5 > scSemi-5: coverage, precision 5/5", + "Beats zero-shot + calibration: coverage 5/5", + "Lower L1 than calibration: ImMicro only" + ], + "footnote": "Completed five-donor comparisons. Multi-visit cohorts used more than five paired samples.", + "footnote_full": "Completed five-donor comparisons. Multi-visit cohorts used more than five paired samples.\nOriginal context: Base f01 6k\u00b71x\u00b7E16 (EMA, DDIM 50) + LoRA r=16, 3,000 steps. LoRA-5 / scSemi-5 = LoRA / scSemiProfiler trained on the same 5 donors (5\u201315 pairs); test = all other usable pairs. Zero-shot = base model without fine-tuning, fed each test pair's real bulk as measured; zero-shot + calibration = the same model fed that bulk after gene-wise calibration, alpha fitted on the same 5 training donors' pairs (both in the same call as LoRA-5). scSemi-5 scored in a separate call, same test set and reference (sha-checked); 25/25 draws. Verdict = better in >= 80% of shared draws. LoRA-5 lower L1 than scSemi-5 in 4/5 (Giroux split 3/5). LoRA-5 > zero-shot + calibration: coverage 5/5, precision 3/5, depth-matched 5/5 and 3/5 cohorts; > zero-shot (real bulk): coverage 5/5, precision 5/5, depth-matched 5/5 and 4/5. Depth-matched P/C vs scSemi-5: same ranking, coverage LoRA-5 0.83\u20130.97 vs scSemi-5 0.15\u20130.28. Composition L1, LoRA-5 minus zero-shot + calibration, per-donor Wilcoxon (gain = p < 0.05): ImMicro -0.03 (p 2e-10), Sepsis +0.02 (p 0.79), Imperial +0.02 (p 0.38), Giroux -0.06 (p 0.38), Ji -0.01 (p 0.70); minus zero-shot (real bulk): ImMicro -0.27 (p 2e-18), Sepsis -0.04 (p 0.15), Imperial -0.35 (p 2e-08), Giroux -0.11 (p 0.04), Ji -0.23 (p 0.02). scANVI labels; P/C = PCA-50, k=10 balls. Scores: scores//fewshot_n5/.", + "sources": "tools/sec09_paired_validation.py; data/sec09_fewpair_cells.csv, data/sec09_scsemi_status.csv, data/sec09_slide_text.json", + "numbers": [ + { + "value": "LoRA-5 vs scSemi-5 coverage (shared-draw means; LoRA-better draws/n; verdict): ImMicro 0.792 vs 0.036 (5/5, lora); Sepsis 0.694 vs 0.080 (5/5, lora); Imperial 0.737 vs 0.059 (5/5, lora); Giroux 0.961 vs 0.607 (5/5, lora); Ji 0.707 vs 0.145 (5/5, lora) -> LoRA verdicts 5/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json and draw_scsemi5/summary.json" + }, + { + "value": "LoRA-5 vs scSemi-5 precision (shared-draw means; LoRA-better draws/n; verdict): ImMicro 0.952 vs 0.518 (5/5, lora); Sepsis 0.906 vs 0.728 (5/5, lora); Imperial 0.853 vs 0.330 (5/5, lora); Giroux 1.000 vs 0.992 (5/5, lora); Ji 0.978 vs 0.840 (5/5, lora) -> LoRA verdicts 5/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json and draw_scsemi5/summary.json" + }, + { + "value": "LoRA-5 vs scSemi-5 comp_l1 (shared-draw means; LoRA-better draws/n; verdict): ImMicro 0.176 vs 0.233 (5/5, lora); Sepsis 0.466 vs 0.612 (5/5, lora); Imperial 0.503 vs 0.602 (5/5, lora); Giroux 0.256 vs 0.270 (3/5, split); Ji 0.513 vs 0.592 (4/5, lora) -> LoRA verdicts 4/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json and draw_scsemi5/summary.json" + }, + { + "value": "LoRA-5 vs scSemi-5 coverage_dm (shared-draw means; LoRA-better draws/n; verdict): ImMicro 0.966 vs 0.284 (5/5, lora); Sepsis 0.834 vs 0.158 (5/5, lora); Imperial 0.863 vs 0.152 (5/5, lora); Giroux 0.974 vs 0.238 (5/5, lora); Ji 0.908 vs 0.150 (5/5, lora) -> LoRA verdicts 5/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json and draw_scsemi5/summary.json" + }, + { + "value": "LoRA-5 vs scSemi-5 precision_dm (shared-draw means; LoRA-better draws/n; verdict): ImMicro 0.975 vs 0.829 (5/5, lora); Sepsis 0.914 vs 0.504 (5/5, lora); Imperial 0.881 vs 0.429 (5/5, lora); Giroux 0.995 vs 0.426 (5/5, lora); Ji 0.974 vs 0.564 (5/5, lora) -> LoRA verdicts 5/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json and draw_scsemi5/summary.json" + }, + { + "value": "LoRA-5 vs zero-shot + calibration coverage (same-draw means; LoRA-better draws/n; verdict): ImMicro 0.792 vs 0.535 (5/5, lora); Sepsis 0.694 vs 0.428 (5/5, lora); Imperial 0.737 vs 0.634 (5/5, lora); Giroux 0.961 vs 0.944 (5/5, lora); Ji 0.707 vs 0.114 (5/5, lora) -> LoRA verdicts 5/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json (arms lora5, zs_cal)" + }, + { + "value": "LoRA-5 vs zero-shot + calibration precision (same-draw means; LoRA-better draws/n; verdict): ImMicro 0.952 vs 0.904 (5/5, lora); Sepsis 0.906 vs 0.899 (3/5, split); Imperial 0.853 vs 0.845 (3/5, split); Giroux 1.000 vs 0.996 (4/5, lora); Ji 0.978 vs 0.782 (5/5, lora) -> LoRA verdicts 3/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json (arms lora5, zs_cal)" + }, + { + "value": "LoRA-5 vs zero-shot + calibration comp_l1 (same-draw means; LoRA-better draws/n; verdict): ImMicro 0.176 vs 0.203 (5/5, lora); Sepsis 0.466 vs 0.446 (2/5, split); Imperial 0.503 vs 0.506 (2/5, split); Giroux 0.256 vs 0.288 (2/5, split); Ji 0.513 vs 0.514 (3/5, split) -> LoRA verdicts 1/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json (arms lora5, zs_cal)" + }, + { + "value": "LoRA-5 vs zero-shot + calibration coverage_dm (same-draw means; LoRA-better draws/n; verdict): ImMicro 0.966 vs 0.780 (5/5, lora); Sepsis 0.834 vs 0.658 (5/5, lora); Imperial 0.863 vs 0.681 (5/5, lora); Giroux 0.974 vs 0.937 (5/5, lora); Ji 0.908 vs 0.357 (5/5, lora) -> LoRA verdicts 5/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json (arms lora5, zs_cal)" + }, + { + "value": "LoRA-5 vs zero-shot + calibration precision_dm (same-draw means; LoRA-better draws/n; verdict): ImMicro 0.975 vs 0.940 (5/5, lora); Sepsis 0.914 vs 0.909 (3/5, split); Imperial 0.881 vs 0.868 (3/5, split); Giroux 0.995 vs 0.989 (5/5, lora); Ji 0.974 vs 0.658 (5/5, lora) -> LoRA verdicts 3/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json (arms lora5, zs_cal)" + }, + { + "value": "LoRA-5 vs zero-shot (real bulk) coverage (same-draw means; LoRA-better draws/n; verdict): ImMicro 0.792 vs 0.190 (5/5, lora); Sepsis 0.694 vs 0.076 (5/5, lora); Imperial 0.737 vs 0.114 (5/5, lora); Giroux 0.961 vs 0.472 (5/5, lora); Ji 0.707 vs 0.066 (5/5, lora) -> LoRA verdicts 5/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json (arms lora5, zs_raw)" + }, + { + "value": "LoRA-5 vs zero-shot (real bulk) precision (same-draw means; LoRA-better draws/n; verdict): ImMicro 0.952 vs 0.677 (5/5, lora); Sepsis 0.906 vs 0.625 (5/5, lora); Imperial 0.853 vs 0.500 (5/5, lora); Giroux 1.000 vs 0.980 (5/5, lora); Ji 0.978 vs 0.799 (5/5, lora) -> LoRA verdicts 5/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json (arms lora5, zs_raw)" + }, + { + "value": "LoRA-5 vs zero-shot (real bulk) comp_l1 (same-draw means; LoRA-better draws/n; verdict): ImMicro 0.176 vs 0.445 (5/5, lora); Sepsis 0.466 vs 0.510 (5/5, lora); Imperial 0.503 vs 0.806 (5/5, lora); Giroux 0.256 vs 0.351 (5/5, lora); Ji 0.513 vs 0.721 (5/5, lora) -> LoRA verdicts 5/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json (arms lora5, zs_raw)" + }, + { + "value": "LoRA-5 vs zero-shot (real bulk) coverage_dm (same-draw means; LoRA-better draws/n; verdict): ImMicro 0.966 vs 0.213 (5/5, lora); Sepsis 0.834 vs 0.073 (5/5, lora); Imperial 0.863 vs 0.151 (5/5, lora); Giroux 0.974 vs 0.549 (5/5, lora); Ji 0.908 vs 0.199 (5/5, lora) -> LoRA verdicts 5/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json (arms lora5, zs_raw)" + }, + { + "value": "LoRA-5 vs zero-shot (real bulk) precision_dm (same-draw means; LoRA-better draws/n; verdict): ImMicro 0.975 vs 0.671 (5/5, lora); Sepsis 0.914 vs 0.595 (5/5, lora); Imperial 0.881 vs 0.561 (5/5, lora); Giroux 0.995 vs 0.996 (2/5, split); Ji 0.974 vs 0.480 (5/5, lora) -> LoRA verdicts 4/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json (arms lora5, zs_raw)" + }, + { + "value": "coverage cohort means over all draws: LoRA-5 ImMicro 0.792, Sepsis 0.694, Imperial 0.737, Giroux 0.961, Ji 0.707; scSemi-5 ImMicro 0.036, Sepsis 0.080, Imperial 0.059, Giroux 0.607, Ji 0.145; zero-shot + calibration ImMicro 0.535, Sepsis 0.428, Imperial 0.634, Giroux 0.944, Ji 0.114; zero-shot ImMicro 0.190, Sepsis 0.076, Imperial 0.114, Giroux 0.472, Ji 0.066", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json and draw_scsemi5/summary.json" + }, + { + "value": "LoRA-5 minus zero-shot + calibration composition L1, per-donor Wilcoxon (draw-averaged): ImMicro -0.027, n=107, p=2.2e-10; Sepsis +0.018, n=82, p=0.79; Imperial +0.016, n=57, p=0.38; Giroux -0.065, n=8, p=0.38; Ji -0.012, n=20, p=0.7 -> gain in 1/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/per_donor.csv; /home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data//folds.csv" + }, + { + "value": "LoRA-5 minus zero-shot (real bulk) composition L1, per-donor Wilcoxon (draw-averaged): ImMicro -0.270, n=107, p=2.2e-18; Sepsis -0.044, n=82, p=0.15; Imperial -0.348, n=57, p=1.6e-08; Giroux -0.107, n=8, p=0.039; Ji -0.232, n=20, p=0.017 -> gain in 4/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/per_donor.csv; /home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data//folds.csv" + }, + { + "value": "scSemiProfiler draws used / LoRA draws: ImMicro 5/5, Sepsis 5/5, Imperial 5/5, Giroux 5/5, Ji 5/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/data/sec09_scsemi_status.csv" + }, + { + "value": "training pairs per draw: ImMicro 5, Sepsis 5, Imperial 12\u201315, Giroux 12\u201314, Ji 7\u20139; test pairs: ImMicro 102, Sepsis 77, Imperial 142\u2013145, Giroux 10\u201312, Ji 24\u201326", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data//fewshot_n5/DRAWS.json (n_train_rows, n_test_rows)" + }, + { + "value": "LoRA r = 16; 3000 fixed steps; base runs/f01_6k_1x_seed20260627/checkpoint_epoch_016.pt; sampling: test donors only, 1,500 cells each, seed 20260927, DDIM 50, EMA, Bernoulli read-out", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/runs/immicro/fewshot_n5/draw0/lora/RUN.json; /home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/README.md" + }, + { + "value": "P/C definition: precision/coverage/density (PCA-50 + k = 10 balls fitted on the scored donors' observed R1 cells); PCA components in every summary.json = 50", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/tools/score_ft.py" + } + ], + "refs": [ + "Kynk\u00e4\u00e4nniemi 2019, NeurIPS", + "Naeem 2020, ICML", + "Wang 2024, Nat Commun" + ], + "refs_full": [ + "Kynk\u00e4\u00e4nniemi et al. Improved precision and recall metric for assessing generative models. NeurIPS 2019", + "Naeem et al. Reliable fidelity and diversity metrics for generative models. ICML 2020", + "Wang et al. scSemiProfiler: advancing large-scale single-cell studies through semi-profiling with deep generative models and active learning. Nat Commun 15:5989 (2024). doi:10.1038/s41467-024-50150-1" + ], + "original_slide": 58, + "figures_before_palette": [ + "figures/sec09_fewpair_cells.png" + ] + }, + { + "kind": "content", + "section": "S", + "section_label": "Supporting", + "text_source": { + "file": "data/sec09_extra_slide_text.json", + "index": 2 + }, + "title": "Precision and coverage across neighbourhood sizes", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/sec09_extra_ksens.png" + ], + "keywords": [ + "Smaller k lowers every set, real included", + "Coverage: LoRA-5 first at all k, 25/25", + "Precision: LoRA-5 first in 20\u201323/25 draws" + ], + "footnote": "Completed sensitivity at k=3,5,10; no new sensitivity campaign required.", + "footnote_full": "Completed sensitivity at k=3,5,10; no new sensitivity campaign required.\nOriginal context: Precision and coverage recomputed with k = 3, 5, 10 nearest-neighbour balls in the scorer's PCA-50 (k = 10 reproduces the main-slide values exactly). Means over the 5 draws per cohort; error bars = SD; grey = real vs real (R1 vs R2, same samples). Both metrics fall with smaller k for every set including the real control (real coverage at k = 3: 0.92\u20130.99), so compare arms with the real row, not with 1. Expected coverage of a perfect generator (Naeem et al. 2020, eq. 9) at k = 3/5/10: 0.88/0.97/1.00 in every cohort. LoRA-5 first on coverage at every k in 25/25 draws; on precision in 23/21/20 of 25 draws (k = 3/5/10); not first: Sepsis 0/1/2, Imperial 2/2/2, Giroux 0/1/1 draws at k = 3/5/10, beaten by zero-shot + calibration. 5-fold pooled design (not drawn): coverage ranking of the 5 arms across k unchanged; LoRA first at every k in 5/5 cohorts. Scores: scores_extra//fewshot_n5/draw/metrics.csv.", + "sources": "tools/sec09_extra_metrics.py; data/sec09_extra_ksens.csv; plans/paired_finetune_intradataset_20260927/scores_extra/", + "numbers": [ + { + "value": "LoRA-5 first on coverage at k = 3, 5 and 10: 25/25 draws", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores_extra//fewshot_n5/draw/{metrics,en}.csv" + }, + { + "value": "LoRA-5 first on precision (draws) at k = 3/5/10: 23/21/20 of 25", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores_extra//fewshot_n5/draw/{metrics,en}.csv" + }, + { + "value": "real vs real coverage at k = 3: ImMicro 0.932, Sepsis 0.925, Imperial 0.919, Giroux 0.993, Ji 0.980", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores_extra//fewshot_n5/draw/{metrics,en}.csv" + }, + { + "value": "every set falls with smaller k (cohort means): True", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores_extra//fewshot_n5/draw/{metrics,en}.csv" + } + ], + "refs": [ + "Naeem 2020, ICML", + "Khayatkhoei 2023, ICML" + ], + "refs_full": [ + "Naeem et al. Reliable fidelity and diversity metrics for generative models. ICML 2020", + "Khayatkhoei & AbdAlmageed. Emergent asymmetry of precision and recall for measuring fidelity and diversity of generative models in high dimensions. ICML 2023" + ], + "original_slide": 60, + "figures_before_palette": [ + "figures/sec09_extra_ksens.png" + ] + }, + { + "kind": "content", + "section": "S", + "section_label": "Supporting", + "text_source": { + "file": "data/sec09_same_pairs_slide_text.json", + "index": 0 + }, + "title": "Same training pairs: five-fold comparison", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/sec09_same_pairs.png" + ], + "keywords": [ + "Coverage, precision: LoRA ahead in 5/5", + "LoRA ahead in composition 4/5, logFC r 3/4", + "19\u201362\u00d7 less GPU time per fold" + ], + "footnote": "Completed original five-fold LoRA/scSemi comparison; not the new total-n curve.", + "sources": "tools/sec09_same_pairs.py; data/sec09_same_pairs.csv; plans/paired_finetune_intradataset_20260927/scsemi/analysis/cv_budget_20261005/", + "say": "Here scSemiProfiler gets exactly the pairs LoRA trains on, every training donor of the fold. LoRA is still ahead on every metric and about 20 times cheaper. Sepsis, Imperial and Giroux are still running.", + "footnote_full": "Completed original five-fold LoRA/scSemi comparison; not the new total-n curve.\nOriginal context: 5-fold CV: for every fold, scSemiProfiler's representatives are exactly LoRA's training + inner-val pairs (the other four folds; checked per fold), test = the fold's samples from their real bulk only. Pooled = one scoring call per method over every usable sample, same donors and reference (sha-checked). scSemiProfiler = upstream c2aea02, default settings, official output. GPU time per fold on one GB10: LoRA = fine-tune + sampling 1,500 cells per test sample; scSemiProfiler = pretraining of every representative + one model per test sample (receipts); data preparation and scoring not counted on either side. scSemiProfiler ahead on composition: Giroux. scSemiProfiler ahead on logFC r: Giroux. Equal pair budget, ImMicro (n = 5, 10, 20, 40 pairs): scSemiProfiler L1 0.218\u20130.227 and coverage 0.04\u20130.04 at every n; LoRA L1 0.171\u20130.188, coverage 0.76\u20130.84 (drawn on the 'Pairs needed: composition, coverage, training steps' slide). Equal pair budget, Sepsis (n = 5, 10, 20, 40 pairs): scSemiProfiler L1 0.516\u20130.536 and coverage 0.10\u20130.10 at every n; LoRA L1 0.365\u20130.428, coverage 0.77\u20130.82 (drawn on the 'Pairs needed: composition, coverage, training steps' slide). Source: plans/paired_finetune_intradataset_20260927/scores//{pooled,scsemi_cv_20261002/pooled}/ and scsemi/analysis/cv_budget_20261005/.", + "numbers": [ + { + "value": "ImMicro: L1 LoRA 0.162 / scSemi 0.217; coverage 0.784 / 0.038; precision 0.924 / 0.505; logFC r nan / nan; top-50 DEG None / None; GPU per fold LoRA 14.7 min / scSemi 4.75 h", + "source": "data/sec09_same_pairs.csv" + }, + { + "value": "Sepsis: L1 LoRA 0.362 / scSemi 0.502; coverage 0.829 / 0.096; precision 0.926 / 0.760; logFC r 0.527 / 0.435; top-50 DEG 23 / 6; GPU per fold LoRA 7.4 min / scSemi 7.69 h", + "source": "data/sec09_same_pairs.csv" + }, + { + "value": "Imperial: L1 LoRA 0.376 / scSemi 0.520; coverage 0.850 / 0.082; precision 0.847 / 0.348; logFC r 0.819 / 0.705; top-50 DEG 143 / 87; GPU per fold LoRA 6.2 min / scSemi 6.18 h", + "source": "data/sec09_same_pairs.csv" + }, + { + "value": "Giroux: L1 LoRA 0.258 / scSemi 0.243; coverage 0.928 / 0.354; precision 0.998 / 0.872; logFC r 0.461 / 0.537; top-50 DEG 5 / 0; GPU per fold LoRA 3.6 min / scSemi 2.89 h", + "source": "data/sec09_same_pairs.csv" + }, + { + "value": "Ji: L1 LoRA 0.504 / scSemi 0.575; coverage 0.857 / 0.167; precision 0.971 / 0.858; logFC r 0.479 / 0.273; top-50 DEG 15 / 0; GPU per fold LoRA 4.4 min / scSemi 1.48 h", + "source": "data/sec09_same_pairs.csv" + }, + { + "value": "Equal pair budget, ImMicro (n = 5, 10, 20, 40 pairs): scSemiProfiler L1 0.218\u20130.227 and coverage 0.04\u20130.04 at every n; LoRA L1 0.171\u20130.188, coverage 0.76\u20130.84 (drawn on the 'Pairs needed: composition, coverage, training steps' slide). Equal pair budget, Sepsis (n = 5, 10, 20, 40 pairs): scSemiProfiler L1 0.516\u20130.536 and coverage 0.10\u20130.10 at every n; LoRA L1 0.365\u20130.428, coverage 0.77\u20130.82 (drawn on the 'Pairs needed: composition, coverage, training steps' slide).", + "source": "data/sec09_same_pairs_budget.csv" + } + ], + "original_slide": 62, + "figures_before_palette": [ + "figures/sec09_same_pairs.png" + ] + }, + { + "kind": "content", + "section": "S", + "section_label": "Supporting", + "layout": "full", + "sources": "tools/sec08_cross_cohort.py; data/sec08_cross_cohort.csv, data/sec08_cross_cohort_steps.csv; plans/paired_crosscohort_20261002/ (README, scores/aggregate.csv)", + "text_source": { + "file": "data/sec08_cross_cohort_slide_text.json", + "index": 0 + }, + "title": "Cross-cohort pairs do not replace local pairs", + "figures": [ + "figures/method_palette_20261013/sec08_cross_cohort.png" + ], + "keywords": [ + "Train on the other cohorts, test the held-out one", + "Calibration ties or beats them on composition in 4/5", + "Own-cohort pairs best on coverage 5/5, disease r 4/4" + ], + "footnote": "Completed leave-one-cohort-out test at K=1000; target-cohort sc was excluded from training.", + "say": "Why one LoRA per cohort: a model trained on the other PBMC cohorts' pairs (three, or four for Ji, plus Liu 2021) rarely beats a zero-pair gene-wise calibration on composition, and no arm without the cohort's own pairs matches the own-cohort LoRA on coverage or disease signal.", + "footnote_full": "Completed leave-one-cohort-out test at K=1000; target-cohort sc was excluded from training.\nOriginal context: Each held-out cohort is scored on all its usable samples against the same frozen reference as its within-cohort benchmark. Zero-shot = base model on real bulk. Calibration from other cohorts = gene-wise alpha (mean pseudobulk CPM / mean bulk CPM) fitted on the training cohorts' pairs, cohort-balanced; 0 target pairs. LoRA r16 / r64 / full fine-tune = trained on the other PBMC cohorts' pairs plus Liu 2021 (training only: its cells are in the pretraining corpus; 44 pairs after excluding 14 hashtag-negative-heavy rows), cohort-balanced batches, inner-validation on 10 % of each training cohort's donors, step-saturation protocol (20k-step blocks, stop when a block lowers the inner-validation loss by < 1 % of the total), value at the last milestone (number in the bar = steps). Ji's bulk is whole blood and is never trained on. Own-cohort LoRA = LoRA r16 on the cohort's own pairs, 5-fold pooled. Giroux's logFC r excludes 196 sex-linked genes. Best on composition among zero-target-pair arms: ImMicro Calibration from other cohorts, Sepsis Calibration from other cohorts, Imperial Full fine-tune, other cohorts, Giroux Calibration from other cohorts, Ji (whole blood) Calibration from other cohorts.", + "numbers": [ + { + "value": "ImMicro: L1 zs 0.446 / xcal 0.172 / lora16sat 0.175 / lora64sat 0.199 / fullsat 0.213 / own 0.162; coverage zs 0.19 / xcal 0.57 / lora16sat 0.45 / lora64sat 0.38 / fullsat 0.35 / own 0.78", + "source": "data/sec08_cross_cohort.csv" + }, + { + "value": "Sepsis: L1 zs 0.509 / xcal 0.444 / lora16sat 0.491 / lora64sat 0.570 / fullsat 0.585 / own 0.362; coverage zs 0.07 / xcal 0.46 / lora16sat 0.34 / lora64sat 0.29 / fullsat 0.35 / own 0.83; logFC r zs -0.02 / xcal 0.10 / lora16sat 0.10 / lora64sat 0.23 / fullsat 0.07 / own 0.53", + "source": "data/sec08_cross_cohort.csv" + }, + { + "value": "Imperial: L1 zs 0.803 / xcal 0.651 / lora16sat 0.635 / lora64sat 0.594 / fullsat 0.580 / own 0.376; coverage zs 0.11 / xcal 0.64 / lora16sat 0.68 / lora64sat 0.72 / fullsat 0.61 / own 0.85; logFC r zs 0.64 / xcal 0.69 / lora16sat 0.49 / lora64sat 0.58 / fullsat 0.46 / own 0.82", + "source": "data/sec08_cross_cohort.csv" + }, + { + "value": "Giroux: L1 zs 0.337 / xcal 0.202 / lora16sat 0.431 / lora64sat 0.410 / fullsat 0.465 / own 0.258; coverage zs 0.29 / xcal 0.89 / lora16sat 0.82 / lora64sat 0.81 / fullsat 0.78 / own 0.93; logFC r zs 0.34 / xcal 0.24 / lora16sat 0.30 / lora64sat 0.28 / fullsat 0.20 / own 0.46", + "source": "data/sec08_cross_cohort.csv" + }, + { + "value": "Ji (whole blood): L1 zs 0.725 / xcal 0.565 / lora16sat 0.740 / lora64sat 0.620 / fullsat 1.030 / own 0.504; coverage zs 0.06 / xcal 0.07 / lora16sat 0.34 / lora64sat 0.40 / fullsat 0.08 / own 0.86; logFC r zs 0.07 / xcal 0.13 / lora16sat 0.05 / lora64sat 0.16 / fullsat 0.14 / own 0.48", + "source": "data/sec08_cross_cohort.csv" + } + ], + "original_slide": 68, + "figures_before_palette": [ + "figures/sec08_cross_cohort.png" + ] + }, + { + "kind": "content", + "section": "S", + "section_label": "Supporting", + "layout": "full", + "sources": "tools/sec08_cross_cohort.py; data/sec08_cross_cohort.csv (minutes = RUN.json wall_seconds / 60), data/sec08_cross_cohort_steps.csv; plans/paired_crosscohort_20261002/ (README, scores/aggregate.csv, runs/hoL_/K1000//RUN.json)", + "text_source": { + "file": "data/sec08_cross_cohort_slide_text.json", + "index": 1 + }, + "title": "Longer cross-cohort training leaves a gap", + "figures": [ + "figures/method_palette_20261013/sec08_cross_cohort_steps.png" + ], + "keywords": [ + "To saturation: r16 20\u201331 min, full 59\u2013107 min", + "Full fine-tune drifts in Sepsis and Ji (Ji L1 0.84 \u2192 1.03)", + "So PBISC fits LoRA on a few pairs per cohort" + ], + "footnote": "Existing saturation experiment; no additional saturation run introduced here.", + "say": "Each run trained until its inner-validation loss saturated. More steps on other cohorts never closed the coverage gap to the cohort's own pairs: Imperial and Giroux improved a little, and full fine-tuning drifted further in Sepsis and Ji (whole-blood bulk).", + "footnote_full": "Existing saturation experiment; no additional saturation run introduced here.\nOriginal context: Same runs as the previous slide. Each point = the milestone checkpoint at the end of a 20k-step block (or at the early stop), sampled and scored on the held-out cohort. Dashed = calibration from other cohorts; dotted = LoRA on the cohort's own pairs. Training stopped (steps): LoRA r16 ImMicro 28k, Sepsis 26k, Imperial 40k, Giroux 35k, Ji 40k; LoRA r64 ImMicro 39k, Sepsis 59k, Imperial 69k, Giroux 69k, Ji 51k; Full fine-tune ImMicro 65k, Sepsis 108k, Imperial 93k, Giroux 120k, Ji 86k. Saturation steps across cohorts: r16 up to 40k, r64 up to 69k, full up to 120k. Training time to saturation (RUN.json wall seconds, one GB10): LoRA r16 ImMicro 21 min, Sepsis 20 min, Imperial 30 min, Giroux 28 min, Ji 31 min; LoRA r64 ImMicro 27 min, Sepsis 44 min, Imperial 54 min, Giroux 53 min, Ji 41 min; Full fine-tune ImMicro 59 min, Sepsis 93 min, Imperial 82 min, Giroux 107 min, Ji 77 min. Gap = no milestone of any mode reaches the own-cohort LoRA's coverage (true in all five cohorts); full fine-tune composition L1 from the first to the last milestone: ImMicro +0.005, Sepsis +0.087, Imperial -0.052, Giroux -0.078, Ji +0.192 (drift >= 0.05 in Sepsis and Ji; improved by >= 0.05 in Imperial and Giroux).", + "numbers": [ + { + "value": "ImMicro LoRA r16: 20k L1 0.163 cov 0.45, 28k L1 0.175 cov 0.45", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "ImMicro LoRA r64: 20k L1 0.196 cov 0.44, 39k L1 0.199 cov 0.38", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "ImMicro Full fine-tune: 20k L1 0.208 cov 0.39, 40k L1 0.211 cov 0.37, 60k L1 0.223 cov 0.35, 65k L1 0.213 cov 0.35", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Sepsis LoRA r16: 20k L1 0.500 cov 0.35, 26k L1 0.491 cov 0.34", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Sepsis LoRA r64: 20k L1 0.550 cov 0.29, 40k L1 0.564 cov 0.29, 59k L1 0.570 cov 0.29", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Sepsis Full fine-tune: 20k L1 0.498 cov 0.33, 40k L1 0.530 cov 0.33, 60k L1 0.546 cov 0.34, 80k L1 0.576 cov 0.34, 100k L1 0.577 cov 0.35, 108k L1 0.585 cov 0.35", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Imperial LoRA r16: 20k L1 0.657 cov 0.68, 40k L1 0.635 cov 0.68", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Imperial LoRA r64: 20k L1 0.622 cov 0.68, 40k L1 0.603 cov 0.71, 60k L1 0.577 cov 0.72, 69k L1 0.594 cov 0.72", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Imperial Full fine-tune: 20k L1 0.632 cov 0.60, 40k L1 0.609 cov 0.60, 60k L1 0.565 cov 0.61, 80k L1 0.576 cov 0.60, 93k L1 0.580 cov 0.61", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Giroux LoRA r16: 20k L1 0.416 cov 0.79, 35k L1 0.431 cov 0.82", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Giroux LoRA r64: 20k L1 0.491 cov 0.80, 40k L1 0.469 cov 0.82, 60k L1 0.430 cov 0.82, 69k L1 0.410 cov 0.81", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Giroux Full fine-tune: 20k L1 0.543 cov 0.75, 40k L1 0.532 cov 0.75, 60k L1 0.499 cov 0.76, 80k L1 0.466 cov 0.76, 100k L1 0.490 cov 0.76, 120k L1 0.465 cov 0.78", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Ji (whole blood) LoRA r16: 20k L1 0.757 cov 0.27, 40k L1 0.740 cov 0.34", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Ji (whole blood) LoRA r64: 20k L1 0.639 cov 0.38, 40k L1 0.668 cov 0.39, 51k L1 0.620 cov 0.40", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Ji (whole blood) Full fine-tune: 20k L1 0.838 cov 0.17, 40k L1 1.006 cov 0.11, 60k L1 1.032 cov 0.08, 80k L1 1.052 cov 0.07, 86k L1 1.030 cov 0.08", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Training time to saturation, LoRA r16: ImMicro 21 min, Sepsis 20 min, Imperial 30 min, Giroux 28 min, Ji 31 min", + "source": "data/sec08_cross_cohort.csv (minutes = RUN.json wall_seconds / 60)" + }, + { + "value": "Training time to saturation, LoRA r64: ImMicro 27 min, Sepsis 44 min, Imperial 54 min, Giroux 53 min, Ji 41 min", + "source": "data/sec08_cross_cohort.csv (minutes = RUN.json wall_seconds / 60)" + }, + { + "value": "Training time to saturation, Full fine-tune: ImMicro 59 min, Sepsis 93 min, Imperial 82 min, Giroux 107 min, Ji 77 min", + "source": "data/sec08_cross_cohort.csv (minutes = RUN.json wall_seconds / 60)" + } + ], + "original_slide": 69, + "figures_before_palette": [ + "figures/sec08_cross_cohort_steps.png" + ] + }, + { + "kind": "content", + "section": "S", + "section_label": "Supporting", + "text_source": { + "file": "data/sec07_no_pretraining_slide_text.json", + "index": 0 + }, + "title": "From-scratch control: evidence and limits", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/sec07_no_pretraining.png" + ], + "keywords": [ + "Same realism; 5 pairs: donor r 0.70 vs 0.17", + "58 pairs: scratch logFC r 0.58 vs 0.49", + "Scratch 61\u201386 min, LoRA 3\u20135 min" + ], + "footnote": "Completed Sepsis fold0 pilot, not all folds. Pretraining supports speed/few-pair adaptation, not universal superiority.", + "sources": "tools/sec07_no_pretraining.py; data/sec07_no_pretraining.csv; plans/paired_finetune_intradataset_20260927/scores/ye2026_sepsis/fold0_scratch_20261005/", + "say": "Pilot on one sepsis fold. Training the same model from scratch on the cohort's pairs gives realistic cells, and with 58 pairs a higher disease logFC r, but with few pairs it cannot place donors: what pretraining buys is the donor-specific composition and minutes instead of an hour per cohort. Folds 1-4 are needed before any claim.", + "footnote_full": "Completed Sepsis fold0 pilot, not all folds. Pretraining supports speed/few-pair adaptation, not universal superiority.\nOriginal context: Pilot on one fold: sepsis fold 0, 17 test donors (12 sepsis, 5 controls); folds 1-4 not run, so no claim yet. From scratch = the same denoiser (54.2 M weights) re-initialised at random (tools/ft_sdd_aug.py --init random; seed 20260627) and trained in full on the cohort's own pairs at the pretraining lr 2e-4, batch 256, EMA 0.999, inner-val selection every 1,000 steps, patience 10, cap 100k steps; the checkpoint supplies only the architecture, 6k panel, value scaler and bulk z-score constants. Pretrained + LoRA = f01 6k 1x E16 + LoRA r16 (lr 5e-4). Same training pairs (5 or 58, + 7-8 inner-val), real-bulk conditions K = 1,000, test donors and sampling seed; recipes = K-diversified pseudobulk augmentation. The 58-pair scratch arms were still improving at the 100k cap. Coverage = PCA-50 k = 10 balls (precision, C2ST, MMD, E_N agree; scores_extra/ye2026_sepsis/fold0_scratch_20261005/). Donor composition r = per cell type (T, B, Mono, NK) Pearson r over the 17 donors of generated vs observed fraction, mean of the four. Dashed = zero-shot. Training time = RUN.json wall seconds on one GB10 (LoRA also needs the one-time 15.9 h pretraining, which every cohort reuses). Source: plans/paired_finetune_intradataset_20260927/scores/ye2026_sepsis/fold0_scratch_20261005/ and README 'No-pretraining ablation'.", + "numbers": [ + { + "value": "5 pairs: pretrained+LoRA L1 0.363 / scratch 0.448; coverage 0.81 / 0.80; logFC r 0.23 / 0.18; donor composition r 0.70 / 0.17; time 3.2 / 61 min; steps (selected/run) 2750/4750 vs 66000/76000", + "source": "data/sec07_no_pretraining.csv" + }, + { + "value": "58 pairs: pretrained+LoRA L1 0.373 / scratch 0.398; coverage 0.86 / 0.86; logFC r 0.49 / 0.58; donor composition r 0.77 / 0.68; time 5.4 / 79 min; steps (selected/run) 6500/8500 vs 100000/100000", + "source": "data/sec07_no_pretraining.csv" + }, + { + "value": "58 + recipes pairs: pretrained+LoRA L1 0.385 / scratch 0.385; coverage 0.87 / 0.88; logFC r 0.45 / 0.54; donor composition r 0.80 / 0.77; time 4.5 / 86 min; steps (selected/run) 3750/5750 vs 100000/100000", + "source": "data/sec07_no_pretraining.csv" + } + ], + "original_slide": 70, + "figures_before_palette": [ + "figures/sec07_no_pretraining.png" + ] + }, + { + "kind": "content", + "section": "A", + "section_label": "Supporting", + "text_source": { + "file": "data/secA_scsemi_readout_slide_text.json", + "index": 0 + }, + "title": "scSemiProfiler readout sensitivity", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/secA_scsemi_readout.png" + ], + "keywords": [ + "Official output zeroes values \u2264 10 (fixed)", + "Removed: scSemi coverage up in 5/5 cohorts", + "No LoRA-5 comparison flips (23/23 same)" + ], + "footnote": "Completed modified-readout check; main results use official output. Models were not retrained for the sensitivity.", + "say": "Appendix only (sensitivity check). The threshold is part of scSemiProfiler's official output, so removing it is a modified run; the main comparison stays the official output. This slide shows that our conclusions do not depend on that read-out.", + "sources": "plans/paired_finetune_intradataset_20260927/scsemi/tools/{resample_output_variants,compare_output_variants}.py; tools/secA_scsemi_readout.py; data/secA_scsemi_readout.csv", + "footnote_full": "Completed modified-readout check; main results use official output. Models were not retrained for the sensitivity.\nOriginal context: Appendix, sensitivity only. scSemiProfiler's official output (upstream commit c2aea02) is the decoder's ZINB mean with every value <= 10 normalized units set to 0; the cut is fixed in the code (vaegan.myVAE.sample, bound = 10, and fast_semi: xsemi * (xsemi > 10)), not an option. At our cohorts' depth (about 1,000-2,000 UMI per cell on the panel) it removes most low- and mid-expressed genes. Here the same trained per-target models were read out again without that cut (CPU, no retraining, target seeds as in the run), exported and scored exactly as the official arm (same 25 five-donor draws, donors, reference; Giroux disease metrics without sex-linked genes). A control re-read with the cut reproduces the official output: pseudobulk r >= 0.996 over 1790 targets, coverage within 0.015. Because this changes the method's official read-out, the main slides keep the official output. Bars = mean of 5 draws, dots = draws. Source: plans/paired_finetune_intradataset_20260927/scsemi/analysis/output_variants_20261003/ (README section 'Read-out sensitivity').", + "numbers": [ + { + "value": "ImMicro: coverage LoRA-5 0.79, official 0.04, no threshold 0.23; precision 0.95 / 0.52 / 0.74; L1 0.176 / 0.233 / 0.199", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scsemi/analysis/output_variants_20261003/per_draw.csv" + }, + { + "value": "Sepsis: coverage LoRA-5 0.69, official 0.08, no threshold 0.18; precision 0.91 / 0.73 / 0.77; L1 0.466 / 0.612 / 0.591; logFC r 0.35 / 0.40 / 0.46; top-50 DEG 16.4 / 5.2 / 14.0", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scsemi/analysis/output_variants_20261003/per_draw.csv" + }, + { + "value": "Imperial: coverage LoRA-5 0.74, official 0.06, no threshold 0.42; precision 0.85 / 0.33 / 0.72; L1 0.503 / 0.602 / 0.603; logFC r 0.75 / 0.56 / 0.66; top-50 DEG 113.6 / 46.8 / 94.6", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scsemi/analysis/output_variants_20261003/per_draw.csv" + }, + { + "value": "Giroux: coverage LoRA-5 0.96, official 0.61, no threshold 0.83; precision 1.00 / 0.99 / 1.00; L1 0.256 / 0.270 / 0.273; logFC r 0.30 / 0.40 / 0.45; top-50 DEG 0.2 / 0.0 / 0.0", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scsemi/analysis/output_variants_20261003/per_draw.csv" + }, + { + "value": "Ji: coverage LoRA-5 0.71, official 0.15, no threshold 0.53; precision 0.98 / 0.84 / 0.92; L1 0.513 / 0.592 / 0.575; logFC r 0.33 / 0.15 / 0.19; top-50 DEG 0.2 / 0.0 / 0.0", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scsemi/analysis/output_variants_20261003/per_draw.csv" + } + ], + "original_slide": 74, + "figures_before_palette": [ + "figures/secA_scsemi_readout.png" + ] + }, + { + "kind": "content", + "section": "4", + "section_label": "Supporting", + "title": "SLE pseudobulk development benchmark", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/sec04_sle_benchmark.png" + ], + "keywords": [ + "42 held-out donors, sc-pseudobulk input", + "Beats scSemiProfiler on all four metrics", + "Beyond bulk: partial r 0.21 vs 0.17" + ], + "footnote": "Completed pseudobulk benchmark; separate from external paired real-bulk validation. PBISC five seeds, scSemi one run.", + "sources": "tools/sec04_sle_benchmark.py; data/sec04_sle_benchmark.csv, data/sec04_sle_numbers.csv", + "footnote_full": "Completed pseudobulk benchmark; separate from external paired real-bulk validation. PBISC five seeds, scSemi one run.\nOriginal context: GSE135779, 42 held-out donors (30 SLE / 12 HC). PBISC f01 6k E16, 5 seeds (mean \u00b1 SD); baselines 1 draw. scSemiProfiler also gets real sc of 14 in-cohort donors (1,500 cells each). Partial r: both logFCs residualised on the input bulk's logFC (the cell-type signal the bulk does not carry), 1,006 genes: PBISC 0.208 \u00b1 0.012, scSemiProfiler 0.167.", + "refs": [ + "Wang 2024, Nat Commun", + "Xu 2021, Mol Syst Biol" + ], + "refs_full": [ + "Wang et al. scSemiProfiler: advancing large-scale single-cell studies through semi-profiling with deep generative models and active learning. Nat Commun 15:5989 (2024). doi:10.1038/s41467-024-50150-1", + "Xu et al. Probabilistic harmonization and annotation of single-cell transcriptomics data with deep generative models. Mol Syst Biol 17:e9620 (2021)" + ], + "original_slide": 42, + "figures_before_palette": [ + "figures/sec04_sle_benchmark.png" + ] + } + ], + "sources": { + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/tools/style.py": { + "sha256": "0d17caf276ae0e9b25285b8fd8f9b75c55621ad3a36382b705a366f327c89de6", + "role": "renderer, style or audited content source" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/tools/fig5_rare_states.py": { + "sha256": "c724d0bb7c204b0363477ea37598c6add5e6cce7fbe2afa72c8f4cc915522213", + "role": "original Fig. 5 cohort-colour mapping" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/tools/results_review_style_plots.py": { + "sha256": "11b6b75a86af474e1c495bc097a1a15ed7e3e7f0635b69b67c0c2f14c51ec8c2", + "role": "renderer, style or audited content source" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores/immicro/fewshot_n5/draw0/per_donor.csv": { + "sha256": "6aae52c2d1a30139025aeae7cc2b44cbecb574f8bd486e2bfbadc02c4aa5444b", + "role": "existing healthy per-donor scores" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores/immicro/fewshot_n5/draw0_scsemi5/per_donor.csv": { + "sha256": "4ae35b3c03aa54dfb0e8ffbc03847b015c9defc6d857e2eae51472ea0f015355", + "role": "existing healthy per-donor scores" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores/immicro/fewshot_n5/draw1/per_donor.csv": { + "sha256": "b988d8608d022898da41c04275110d95f0b28daa145bec10027ed7e9193cd473", + "role": "existing healthy per-donor scores" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores/immicro/fewshot_n5/draw1_scsemi5/per_donor.csv": { + "sha256": "29e5d70b1c15ad2d850dd3cc805b32f693b415c75507920f9ca87e2c4c55f7d0", + "role": "existing healthy per-donor scores" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores/immicro/fewshot_n5/draw2/per_donor.csv": { + "sha256": "1caf3d114796869aa6a0fb6edbeddc93dfc8efe8dbd7bd332cb9ea12f0d1a79f", + "role": "existing healthy per-donor scores" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores/immicro/fewshot_n5/draw2_scsemi5/per_donor.csv": { + "sha256": "0f257a24ef9dde20e4f1ec8dce1bbcc7199a5ec01d8efa37de27f32d5689eef2", + "role": "existing healthy per-donor scores" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores/immicro/fewshot_n5/draw3/per_donor.csv": { + "sha256": "a9bf586d886d1530712b53ae01457b20fb2b9018e698d4bfcb3b6c80a883e5d7", + "role": "existing healthy per-donor scores" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores/immicro/fewshot_n5/draw3_scsemi5/per_donor.csv": { + "sha256": "9f9d6a9777d160895e4cc102e496cbf385a8713e95e871a77ebb3165779cbe28", + "role": "existing healthy per-donor scores" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores/immicro/fewshot_n5/draw4/per_donor.csv": { + "sha256": "c6bd999a3c963f9ef7f5dc8deeb5a7255c5dae709b4d7fdb330df1efba4c8e8f", + "role": "existing healthy per-donor scores" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores/immicro/fewshot_n5/draw4_scsemi5/per_donor.csv": { + "sha256": "43ae3ef666fe05f31f023fb8bd0af132478aea67d2f3fb4e56918c4c1b9242fe", + "role": "existing healthy per-donor scores" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/data/composition_error_20261013/cell_type_summary.csv": { + "sha256": "a073a04ccb4ec8e8c7b3398a758a708261b64dcb1dcf141483551fcdb723ff54", + "role": "derived donor-weighted absolute-error summary" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/data/composition_error_20261013/paired_donor_errors.csv": { + "sha256": "a908315a22b13c0821550562937e1308225df08fb2535f06f079e52446b1a258", + "role": "derived paired donor errors; repeated draws averaged within donor" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paper_figures_strictpairs_20261012/source_data/rare_all_states.csv": { + "sha256": "9dc216b4172ff9e46ac0c498f52049ac1869df742761208cc48597ad3397b36f", + "role": "existing corrected rare-state summaries" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paper_figures_strictpairs_20261012/source_data/rare_all_states_aggregate.csv": { + "sha256": "5fd9a2a4051ee6443ed20ec26db2499f9bf7b89116e3a373abc1bfacd8737317", + "role": "existing zero-aware rare-state metrics" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_healthy_composition.png": { + "sha256": "6c338b19e27786ea980609a3df9202a3fa9a9eb638c31a410c8cf696a8375b70", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_healthy_composition.pdf": { + "sha256": "358e6c5226f43506d7424147762eb92f40ae68811f8a2732ba02dcd7711c7123", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_healthy_composition.svg": { + "sha256": "7c23124486da965d45740f6cb21bbe87e447afdd675326d606bf5120d9cd1de5", + "role": "restyled saved-table plot" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_rare_observed.png": { + "sha256": "4bf0cb00b3bdf40bee11ac9957ae12e176da79e0da8be2a3e9790b2accb91a6f", + "role": "restyled figure" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_rare_observed.pdf": { + "sha256": "1176fa69f159dcdb1a06ebd2ad6b4753675db62ff4622355bde3f0a5439f3b87", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_rare_observed.svg": { + "sha256": "f32450216f769baa0a6b28da0fc09a26f3fc64071593d2e417931db43b35b27a", + "role": "restyled saved-table plot" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_rare_corrected_comparison.png": { + "sha256": "872fa269a413be4357689f3ef2a7029a43ea29909463797f8d15c7e8ead08217", + "role": "restyled figure" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_rare_corrected_comparison.pdf": { + "sha256": "dc1f37541f1bc3fb9628997abf1d14718b983f190703319819c4ecbd73f0c763", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_rare_corrected_comparison.svg": { + "sha256": "70212b3a0b27827369cfce593f4dd00a3e00ae6e42b02ea99e9324d2573ac6f4", + "role": "restyled saved-table plot" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_ablation.png": { + "sha256": "da9d41847d5768628c5954656bf95d012e16485c18403cbf247a98cc361ce361", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_ablation.pdf": { + "sha256": "f11a45f022ec55957823d1fded3589733d8cb51083857a59d45140abcedd057b", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_ablation.svg": { + "sha256": "e3b5ef57548ab96355388b4905dd4b7a89cb2bb3791c8723cc2057e30069a91c", + "role": "restyled panel svg" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_ablation.provenance.json": { + "sha256": "1135791a30b1720f5ec934a4aaa5c077fb1ddc8cfa081408d55bcaee63347281", + "role": "restyled panel provenance" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paper_figures_strictpairs_20261012/source_data/ablation_pretrain_pc.csv": { + "sha256": "de8f99d2c489a809e16bacb07492f7986f552b10e19dec063eb44c06077cdaa2", + "role": "saved panel data" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paper_figures_strictpairs_20261012/source_data/ablation_pretrain_pc_receipt.json": { + "sha256": "d6b96cdc5303bdce7eee408a18e1bb21f7022b501df1884861904df245286ff3", + "role": "saved panel data" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paper_figures_strictpairs_20261012/source_data/ablation_pretrain_composition.csv": { + "sha256": "d6654b07553780a9387bdfd2cee3ab6979f24c3b7ab9d6f02678bcc974f5482b", + "role": "saved panel data" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paper_figures_strictpairs_20261012/source_data/ablation_pretrain_composition_receipt.json": { + "sha256": "a5d2c28ca120a1f2788932b888a66329a68d4955960f31cd89f4f6de6b72607a", + "role": "saved panel data" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_rank.png": { + "sha256": "bdba0e29c46e69aa395eeff31ce42bf180dd7271c797a99382531c156cd31923", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_rank.pdf": { + "sha256": "f335bced08725f02a266e8130391d623ae61a2ce0c236a716e27499125755bff", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_rank.svg": { + "sha256": "f4e3b43efe99114634bb77e317fd48e1a5772a9a00622141701823ec018b7066", + "role": "restyled panel svg" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_rank.provenance.json": { + "sha256": "39e9ceebf7d357645d835e845307fd18dab39efcc171d512b6edce1e4b2503aa", + "role": "restyled panel provenance" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paper_figures_strictpairs_20261012/source_data/fig2_historical_rank_full.csv": { + "sha256": "5ebed262f69a6c4cac932d7bb81dc325665b11e1844fc0e92fdd3f4530ab5ef0", + "role": "saved panel data" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paper_figures_strictpairs_20261012/source_data/fig2_historical_rank_full_provenance.json": { + "sha256": "65913566e58c8587815583eef0aeff8c3070b75eceee90725540e324883aacc8", + "role": "saved panel data" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_behcet.png": { + "sha256": "692ecf5ac216c661f465cb9b534e5809978fdfcd8dfc82376b7d1517093ee9b0", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_behcet.pdf": { + "sha256": "e775eb8186e929bc2598b9e758c0f88b6142510b4a1d4a800752c799a6f6e095", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_behcet.svg": { + "sha256": "599d3c587640f9787bf7288577a860ff3cfbdbc42ff43ef3bcc77ff6feb48e76", + "role": "restyled panel svg" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_behcet.provenance.json": { + "sha256": "f17fdaec4c7c25547bdab379bc0ca28fda5388f96d17013a0682ad281b1553c7", + "role": "restyled panel provenance" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paper_figures_strictpairs_20261012/source_data/rare_behcet_per_sample.csv": { + "sha256": "d19b5007de6861ba30730c4e15c90c78bf4d5e1f115ea9464ac66adbe5e2722e", + "role": "saved panel data" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paper_figures_strictpairs_20261012/reports/rare_behcet_provenance.json": { + "sha256": "f2147fb96c914cd957853b151191fec004079066dd7ab017e16b508512f6554f", + "role": "saved panel data" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/data/method_palette_20261013/palette.json": { + "sha256": "9482022c4e49548c1333e3be0fb2a3063fb9a923ca1c0dfd73af3712776d5da8", + "role": "user-approved method colour policy" + }, + "/home/sj_server_1/PBISC/paper-release-20260928/manuscript/figures/fig01_architecture_nature.pdf": { + "sha256": "95ccf98eb7048375505fb59a074a1a870a4d71ea4a8263b2b283e10a80883b90", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/fig1b_overview_nature.png": { + "sha256": "e50ff68c0a08b0c4730c7a5abf2700cd1b994afc8ca2e69de60d33ff99f8c9e1", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/fig1b_overview_nature.pdf": { + "sha256": "c484f7ebc0ea830a7211807749ea4bfbad8fd5ad02d507526617d8bcb4b111f7", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/fig1b_overview_nature.png": { + "sha256": "9f44801a6172013437da48224361cf02167445735c83254cb50170b636c48fbb", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/fig1b_overview_nature.palette.json": { + "sha256": "957acec55816e43e8d6c6509d328798c789059554d2d4349af8798d01c0d53ae", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/review_style_20261013_ablation.pdf": { + "sha256": "2c050bce8f14bb689093671b41de20a4d97f90a62797378cb1ad217fbf4ce62e", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/review_style_20261013_ablation.png": { + "sha256": "619fea03e7931592991cff1f15ad8fae50085ab8880d88d8bc033647206a783b", + "role": "restyled figure" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/review_style_20261013_ablation.palette.json": { + "sha256": "4ea79585e9eae48e715436d5ca603d92629a6b7f9cc4e53cafdbf6fb03ffdb01", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec03_panel.pdf": { + "sha256": "ddbfdbab9cfc924ab7f1f6a3f5c59a74c17648166b329986bb8a075c5eda2986", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec03_panel.png": { + "sha256": "de0a00095554166c61901b9fa5c8dee3a813d2552620baa88147251d5b2600a2", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec03_panel.pdf": { + "sha256": "5bc01e2094d79f5dbd7703f91a569104ef4008bd7850ead0ee12f2ebbbda238e", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec03_panel.png": { + "sha256": "fbc218acf1a3e714ed79ec452fa096cb7bd35339bf8e9f06bf5883e9e3d75c7c", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec03_panel.palette.json": { + "sha256": "0af8258d339abb4c2172be8e622bfac368048f9256d98e71be79ab434b587d4d", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/review_style_20261013_rank.pdf": { + "sha256": "07f3f3bbdcb441810379d1f25acfe4b769f44f2d2ae319399e784d1507d29d98", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/review_style_20261013_rank.png": { + "sha256": "a86136bc6b24cc8da4a277e71a08d4a4b68d70997363ad21c5876861917b0d67", + "role": "restyled figure" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/review_style_20261013_rank.palette.json": { + "sha256": "2a4e3ee099e3c4814f673ac798b13ddfe30ded5e88f524131a0fb21ece49288e", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/fig3_healthy_pairs.pdf": { + "sha256": "37ea09fc9802cb5d799ca5b3db54d9a36f3eec43ba023563952f844b730d8802", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/fig3_healthy_pairs.png": { + "sha256": "cb85e602f4f94592a5939ec32fc07ecafad041ceb7093f161f7d84490ebe427b", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/fig3_healthy_pairs.pdf": { + "sha256": "c3b302515b2af96a4a323513c1056ed7c7e8229c13c6054688d60cd764561998", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/fig3_healthy_pairs.png": { + "sha256": "07029e95bc657d5e99f42aa629d663ff7cee97ff166ccf549b1cfb9a1a5b7781", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/fig3_healthy_pairs.palette.json": { + "sha256": "caab5dc7ca8c56b04a1a9aa0444be3257f732265f55ddbc603bbd539594d7d9d", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/fig2_healthy_pairs.pdf": { + "sha256": "78a83d2603464842b32553814d1cfa665383b74f7e19438ea03440d6719c0db2", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/fig2_healthy_pairs.png": { + "sha256": "d3d9156da0907953644a8bfce47dbbd547f64b49f65c183332007fa49bf19aae", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/fig2_healthy_pairs.pdf": { + "sha256": "cfb77dbd5672629e3bd05035898891e06bfb3535de074f2677fbfa9c2058fe16", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/fig2_healthy_pairs.png": { + "sha256": "c3d8cf7cb41f0c29bd6521d2743ffe6a319c449d28dd6371d9edcb02fd9213b2", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/fig2_healthy_pairs.palette.json": { + "sha256": "5c70ab54aa11c49e66360b9651cb018f8fea2a9651c09b8e451968930f9d3cf8", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/fig3_healthy_fewpair.pdf": { + "sha256": "d93b8cc9c49be301038c021c00e6d41af87aae06ddeeebca2fb8092a2fe808f1", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/fig3_healthy_fewpair.png": { + "sha256": "887de722878a31776bf550b4cddf8ab3b9a064ac6cb5f1d0a49ac2cb2b0c7256", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/fig3_healthy_fewpair.pdf": { + "sha256": "345bd8b9bbddeadc3a71ac1591f4a7ebeaf212ecd0fde5393bd0c0f7c9fd611d", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/fig3_healthy_fewpair.png": { + "sha256": "61da8d0eb214b96e807c4057cadbbfeb1bec31af20afeab4ce981d9b445515a3", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/fig3_healthy_fewpair.palette.json": { + "sha256": "f3a46cf767ef63a3ad71c88c078783e101d4be28821ff158aa89a7b899084bfe", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/review_style_20261013_healthy_composition.pdf": { + "sha256": "d4ef5757207238ebb5618a09681e04eeb9bee8b3bf088536042f5bae40d3a422", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/review_style_20261013_healthy_composition.png": { + "sha256": "b57ad20310ec5f162bc6c9e3cc07b7fdb3534dcd4c638d9fa07dd2d04b21095e", + "role": "restyled figure" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/review_style_20261013_healthy_composition.palette.json": { + "sha256": "cc2d0a890456cb7be18edbd7e826fbb0cf4811d5460414ef7b953604cefebbc2", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/fig3_healthy_copy.pdf": { + "sha256": "644c45e3a49eb7d16b15c4da1736ce5eb94ea8d2dd0213ce820a6d111b215964", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/fig3_healthy_copy.png": { + "sha256": "c8c01fe00bc5f4fd259b32cee19f0a166984671a59900bf8a9d5e942568693d9", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/fig3_healthy_copy.pdf": { + "sha256": "9084f439ed06c80249770240455de7070b1e8f105db8b9e247726197e95de188", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/fig3_healthy_copy.png": { + "sha256": "951f915cc19535c6f5d2b1085358cd1d14b3bad2b671d2a808731784886fb6d4", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/fig3_healthy_copy.palette.json": { + "sha256": "ab3fe55c03ef458517fefb61191afff408bcc0595854145dc66244a2e022ecc4", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec09_time_to_apply.pdf": { + "sha256": "614303f5f76d1ce103b93ebb38973c93c2463e20d78055cd430f8fea2d84929a", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec09_time_to_apply.png": { + "sha256": "c6e6bac98c41cc9c807686565cc461eee78a35fc645da8c64c5d7a36267c955b", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec09_time_to_apply.pdf": { + "sha256": "45d5f850e7181a0544e425ee735211dbbf2669cfb3ffb795b07cc4a70265da91", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec09_time_to_apply.png": { + "sha256": "a9a9c0e6fbdde7f6b9043550052fbe6ab28e86178b364e9d05828f2d00ccce24", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec09_time_to_apply.palette.json": { + "sha256": "6addd32748f29af88e8dfdc3c2f82c17e67dbfaf506b02203c1882fe4a5ce1ad", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/fig4_disease_fewpair.pdf": { + "sha256": "b7c727cf4c76755c4234f07a22f5f314000aa762982d141eb4f6ac60c802bfb6", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/fig4_disease_fewpair.png": { + "sha256": "5791a938059787ada171323091224f753ca3f5d31ef82d4cb7460cccf8316fa0", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/fig4_disease_fewpair.pdf": { + "sha256": "4baf592a2a049488932fd476af4084897b8647e285c47a523b9b4992d80359f2", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/fig4_disease_fewpair.png": { + "sha256": "8247ac287586f6cdfafb25ce5e9f7cb6d54a2970f7101e7643c9a5b25c10dba4", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/fig4_disease_fewpair.palette.json": { + "sha256": "b73318fa454151913dea71b9b75cdaa3f7048f72e27dc3b3f295d43847436614", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/fig4_disease_fivefold.pdf": { + "sha256": "003d005069e886f31855cbf455b8f287bf59fe3532ee1cf95826888961689fb7", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/fig4_disease_fivefold.png": { + "sha256": "22ff8bef1f93b9dda2b42ab53b3380e777bdd6f25cf3c5c928fc952f8201d1d4", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/fig4_disease_fivefold.pdf": { + "sha256": "ca7328676d963c9e88ba2dfbc4fdff3a7d76061e7aedcc4b529653be42953722", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/fig4_disease_fivefold.png": { + "sha256": "2fd600d7b471c3a405c653e6f270a32fe9b35e7b0e77e6bcceabace9a4ff3a41", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/fig4_disease_fivefold.palette.json": { + "sha256": "0074c07c18c847d3b5247168fae4bb8850532db9acaf83f324e4b56a2c789ef9", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec09_cat_curves.pdf": { + "sha256": "09f3bbd11ca995584b0015800e2afbe02c3fa080efac52851f836f0d956ab8f2", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec09_cat_curves.png": { + "sha256": "cf72097c3b6432105bbcc12e339a3f6e27fd916f4ad613a0c4a644a8c82d9ebf", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec09_cat_curves.pdf": { + "sha256": "9f48ae8f448e0bd5bf1d6d2d69702dcad1f28e3891093f0c69a6526b2c4be7cb", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec09_cat_curves.png": { + "sha256": "059a8c1b31ae19e904231d3ee00193b04d6e51a76dc192b4c0dbb3445d3771c1", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec09_cat_curves.palette.json": { + "sha256": "0af590bca675a6f690fab0cbde0a9776ef1ec7d069ea5d376b4bca490a0d16bf", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/secS_deconv_disease.pdf": { + "sha256": "31ddaf4d792943114ae52c07a330a3f63d636b8adca9ca311aa6b0449a812f80", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/secS_deconv_disease.png": { + "sha256": "4439a3555b5f869346bf412554a32a286b70e80c190383b5765c8c212b29f06f", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/secS_deconv_disease.pdf": { + "sha256": "516eabb18afb6f990564510137ce2b34748b3000fbcb34901cb086ac34855c6b", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/secS_deconv_disease.png": { + "sha256": "4942a321a803707f45d290dd0464cac6268e740e0168162e6e637cae17d5f7ad", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/secS_deconv_disease.palette.json": { + "sha256": "9669af3c51f4594766f3c60c76d7f7ad6451962695aa5dea883f1ae3b436326f", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/secS_deconv_composition.pdf": { + "sha256": "54a24da70bf18e208b459775586cdc75643d2bd585526c8e92875e415cf811e7", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/secS_deconv_composition.png": { + "sha256": "e48667baef7b45ca34988455d13c28039cfd7c4a292d779d1619424c1e0660ed", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/secS_deconv_composition.pdf": { + "sha256": "aa2de99e29d57d0dbbd70924c8aad0bdde51f306790b4cfa837d42269237cd82", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/secS_deconv_composition.png": { + "sha256": "64140caec57d4d2bd0e63eaf5b94865940699a0d2cee093199cf3ed10bb0633e", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/secS_deconv_composition.palette.json": { + "sha256": "601a55720318b82e182a9ae4f7d9d9515ab6fa8c3f74af065be2943fb9246033", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/review_style_20261013_behcet.pdf": { + "sha256": "3ebaec95121c1527855aacf35f77c0a4116edfc74f372f35b9c355be2e236380", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/review_style_20261013_behcet.png": { + "sha256": "455d5bdea5cdbf15ecf3adb05aeb9f281207d2d1c1eacc147c16e7040aa808d8", + "role": "restyled figure" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/review_style_20261013_behcet.palette.json": { + "sha256": "680d37a05fd11def50cada34f6e7dd209fae24dcb16b4caeb0fb69b1258123d2", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec05_cells_metrics.pdf": { + "sha256": "a555f4bced0da2cc6ea25d35c27b0a4e0992c377f40abc356edda85e9e2fed92", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec05_cells_metrics.png": { + "sha256": "775ef952401906945d79ebc0a4e0f89ea693c0598e293d9d3739efc0ab0cd6b9", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec05_cells_metrics.pdf": { + "sha256": "73975cfde382668a4d5c2fbf988ee270a649c6f6c6a6c881dc11d23c8bfd881e", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec05_cells_metrics.png": { + "sha256": "0622d83bb9f5415fc2e1b37ea7d43415c64cb067f6b588780dd49fed67482ecf", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec05_cells_metrics.palette.json": { + "sha256": "90c61ee8f136e4ae311ba0a642e3e1b83aacbc5b12c0aca208cbfe0b74d4b027", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec09_fewpair_cells.pdf": { + "sha256": "ed9849628f3b1556846bef6359aa09ffd79983541049bc518a74187bac916e21", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec09_fewpair_cells.png": { + "sha256": "fe491dd0a6d9eda2da5ee31ad623b8299e812f0205921cfedba557867ab2f0b5", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec09_fewpair_cells.pdf": { + "sha256": "6cc23c842f33b93f23bc779b30c6ccbe2dfd8dc72b12a36410ad66de7929ad3d", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec09_fewpair_cells.png": { + "sha256": "4ed33d98fb653c14f054070212c2ce463a862881d600902f704353234a151863", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec09_fewpair_cells.palette.json": { + "sha256": "ee462ac5c03fb0c4ebcac2ec9cdd837fd3c49dfe441dac8e33beb5b697b29718", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec09_extra_ksens.pdf": { + "sha256": "a3ef5e661ccfff24e5ee0e25b5a6e33c9c34f6bc3e41aac0d222b319a87413c1", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec09_extra_ksens.png": { + "sha256": "9a915c33ad25de3d411dd0a318b7efa68815515aab64dd31b6716c5bb1b14209", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec09_extra_ksens.pdf": { + "sha256": "3072e860508522360c2b1b6ea19ce22dd9faae6e13a5110ecd6da6d4f5caa6b6", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec09_extra_ksens.png": { + "sha256": "6420a328dbe741202adff890b4ea588fcb92f0313222366d3a950d42e9de4015", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec09_extra_ksens.palette.json": { + "sha256": "2ff5d3bf49b154885c8f8ecc6b03b054de357eebe6222a1e8f80837654f33a7a", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec09_same_pairs.pdf": { + "sha256": "85e1244526ce3967fad550ba47ed9a121fb91bcd3798c8fb42d0ab09d5d989b4", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec09_same_pairs.png": { + "sha256": "41908506775df46eae5796bd83469a9023677d92857f8714abc4319e361c5ee2", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec09_same_pairs.pdf": { + "sha256": "71eb528a31f89ed0fb4196a6ac6b233557c1d37e09bf5864ff8c6666a751a71c", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec09_same_pairs.png": { + "sha256": "b33fc9ddaf90129ec340656b5064ab1a1bd774c24c85f88d6aacade83df6ddea", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec09_same_pairs.palette.json": { + "sha256": "5fde31527ee27437ea0303bff2590c9ce5cca7d3d9a0e44ed417fa0d3ef9f361", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec08_cross_cohort.pdf": { + "sha256": "cae7ea9560a094324f8a71105524a5d72b5977168c9478b52649c8fd3338445c", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec08_cross_cohort.png": { + "sha256": "9d979e00ba7abe1f5ca8b09ad0e72e9bd84415cd5e5f1d9f55e3e58c5968f8b3", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec08_cross_cohort.pdf": { + "sha256": "52af9666df540e4647a6349643776fc1b86f221a6f77257b2051f734221a9750", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec08_cross_cohort.png": { + "sha256": "7d48bfb3b1453fd271e91ad78d0b06255a34078c8f96ee87e459acda4999bab2", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec08_cross_cohort.palette.json": { + "sha256": "4f1334bd33573c3609e4fbf07c4ee199b4a62ba3880cc54adefcefdbd0867b61", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec08_cross_cohort_steps.pdf": { + "sha256": "ac0e1b49613dba0f44c1bfbfbb7c964998ee4ff5782f686fa0f4162f9634da33", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec08_cross_cohort_steps.png": { + "sha256": "13c926b355024c9f844acc41e453a1bae5140bf45252c1b199a55c25c218d7ac", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec08_cross_cohort_steps.pdf": { + "sha256": "63cf8b73686667f6a665dbc56911870cfd0e42e645f84544054bf93bd939a1af", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec08_cross_cohort_steps.png": { + "sha256": "bc089c6d80383fe2fdb50fd50d35602f4582cee8a57c83b937813babdc20bd8d", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec08_cross_cohort_steps.palette.json": { + "sha256": "c7f670a9543c85f7b235f48abd84a61b0551f83aa7de11e7cac8e95e618ec6c3", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec07_no_pretraining.pdf": { + "sha256": "4a46aa5e38a600d2f67e051e165b1ffdc1afb6c564b2352a9d58eb5e5e4f0f7b", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec07_no_pretraining.png": { + "sha256": "a9d51b57a7bcfb479df28b7d4b3d58ecec103d81fd30223789b1b772cc06952a", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec07_no_pretraining.pdf": { + "sha256": "d75474558a6d8ffaea3a151f81ba3828a927622850230735e95d0fb01e0153a2", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec07_no_pretraining.png": { + "sha256": "80a88ce1deae90db43c95332120415641279f644b3fd8ce2f477e3fc09272123", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec07_no_pretraining.palette.json": { + "sha256": "23ddf4204373f1e0c629dc26445bcd18b02386ee64c12b4e5ff2ad8352cba233", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/secA_scsemi_readout.pdf": { + "sha256": "8d88d3ba4d9e6bab62c739d03830dd5c90a72a580ea6ac1bca679ab6584ccad2", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/secA_scsemi_readout.png": { + "sha256": "06468c2e50e6107184e5345dbfe3e9e79783410b3ff30f133f84a7e28d0d5b8b", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/secA_scsemi_readout.pdf": { + "sha256": "59138d144773dbb9b1370584776fe5483db53f012362ba8fae39df89b30fc418", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/secA_scsemi_readout.png": { + "sha256": "af78838b41ab4e17823dc4cf33ed9503add68978ab6350d5f9bf97b302e42a9e", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/secA_scsemi_readout.palette.json": { + "sha256": "be8f2dfcc50b9d3579f4125c94cc954aa32fa427c55c9e25a046b4132892525b", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec04_sle_benchmark.pdf": { + "sha256": "17d1cbaf83f9e5f764ab61e445a8d9d10f5f5434b5abefae5dedf2d27210e292", + "role": "original vector figure for method recolouring" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec04_sle_benchmark.png": { + "sha256": "3e79ca869e511ede1c7b438c3f0b191d2991b89ab3e6763f110bcf3f84069592", + "role": "original figure image" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec04_sle_benchmark.pdf": { + "sha256": "de7de63a5d3b0c08060efb7362384ac5268945cc8a5c25cef2d7cb7a23cf2b53", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec04_sle_benchmark.png": { + "sha256": "c916d7433829b927c60751f76aa26ca7eb9cdaceecb2c360b0bd5ffd3b2c86c8", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/method_palette_20261013/sec04_sle_benchmark.palette.json": { + "sha256": "4844ba1d4afa5be5d38f90fd91f9d7dde0335f9def4e1995f58ca4615f185893", + "role": "method palette variant or preservation proof" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/data/method_palette_20261013/ASSET_MAP.json": { + "sha256": "0c53e5fb122c3f7445b003e3505c772f815852aa56b36098fc3b6afa54c3dcaf", + "role": "method palette coverage and per-figure proof locations" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec02_bulk_conditioning.png": { + "sha256": "3ff1ec63ef0132d62b2e8686f1befa5f002e86f3793868e015efa4087648e399", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec02_bulk_conditioning.pdf": { + "sha256": "75ed029d4acc720fdbf85203e7cb0a65b32fc88ba26ba8652d4f686d1e0b1d58", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/fig1d_corpus.png": { + "sha256": "5f377103a33cf55e9da7e579db06214262360ba75ec5a8e586ae3984a0961513", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec05_condition_space.png": { + "sha256": "474de533f201f7da4e5aa9f2712d79cb9d58011888db3c830f5bf216d7b613c3", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec05_condition_space.pdf": { + "sha256": "e0701bd49471a731a4ec6489770a7187cda28727c8a176ab681ad9381624f4c2", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec09_cohorts.png": { + "sha256": "4624a15d5984a0883adc5095fe288f52e543bf04ee1986ca07adbb46254872c0", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec09_cohorts.pdf": { + "sha256": "948538720b7f3be73cab30e3e26819ab9c46021c1d65bab1d112eb0a3afac88d", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec05_cells_umap.png": { + "sha256": "f752764621e0f92a65e1e9f9901074339e1d3a241e73c9ab3990065a0d2fae9d", + "role": "figure reused intact" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/sec05_cells_umap.pdf": { + "sha256": "137e1d1351e9dc79aa3398590468d53835ef4341fd5d64dc29c5ebb02d2dfaed", + "role": "figure vector companion" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/generate_deck.py": { + "sha256": "005bec4082f16a095b2db207ec02fa8155226825ea2e243b406ed0a2108fa180", + "role": "renderer, style or audited content source" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/data/slides_spec.json": { + "sha256": "6b8fa055e20a3379ce2e50ca19dbbe285d81e4e95d2c42597e236945b8b2bc23", + "role": "renderer, style or audited content source" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/tools/build_results_review_20261013.py": { + "sha256": "1aa8fee9e38d2c95ab785d0754175370b6420d116e1d56c36a9c1c3701bf0f7d", + "role": "renderer, style or audited content source" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/tools/results_review_style_panels.py": { + "sha256": "395662295dd314369711123ecadda7ec5faa7cc916b55cdf2ba3f55307cc56cd", + "role": "renderer, style or audited content source" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/data/style_review_20261013/before_results_review_20261013_provenance.json": { + "sha256": "76f8f05de6cb82240e5c40fe85826d28822c436ec0bc2dae07fe5a68cd91ef79", + "role": "renderer, style or audited content source" + }, + "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/tools/review_palette_pdf.py": { + "sha256": "c5b6280defb4ee282f829a2cd4e79efed59abf63ff55dd14df0505c5d074320e", + "role": "renderer, style or audited content source" + } + }, + "new_training": false, + "new_generation": false, + "cpu_work": "Saved CSV plots; healthy composition now compares paired absolute errors and donor wins. No new model outputs.", + "plot_calculations": { + "healthy_composition": { + "estimator": "100*abs(generated-observed) per donor-draw; average held-out draws within donor; equal donor weight", + "comparison": "paired errors, not error after averaging proportions", + "rows_per_arm": 510, + "donors": 107, + "cell_types": [ + "T cell", + "B cell", + "Monocyte", + "NK cell", + "DC", + "Other" + ], + "summary": [ + { + "cell_type": "T cell", + "pbisc_mae_pp": 6.219426350450708, + "scsemi_mae_pp": 6.862128730026533, + "scsemi_minus_pbisc_pp": 0.6427023795758255, + "pbisc_wins": 63, + "scsemi_wins": 44, + "ties": 0, + "donors": 107 + }, + { + "cell_type": "B cell", + "pbisc_mae_pp": 1.9524158913161018, + "scsemi_mae_pp": 3.174458115864143, + "scsemi_minus_pbisc_pp": 1.222042224548042, + "pbisc_wins": 93, + "scsemi_wins": 14, + "ties": 0, + "donors": 107 + }, + { + "cell_type": "Monocyte", + "pbisc_mae_pp": 4.550689595899203, + "scsemi_mae_pp": 5.389708273059629, + "scsemi_minus_pbisc_pp": 0.839018677160426, + "pbisc_wins": 61, + "scsemi_wins": 46, + "ties": 0, + "donors": 107 + }, + { + "cell_type": "NK cell", + "pbisc_mae_pp": 3.638846944443304, + "scsemi_mae_pp": 4.415005942815469, + "scsemi_minus_pbisc_pp": 0.7761589983721648, + "pbisc_wins": 78, + "scsemi_wins": 29, + "ties": 0, + "donors": 107 + }, + { + "cell_type": "DC", + "pbisc_mae_pp": 0.8118366446019785, + "scsemi_mae_pp": 3.023393579077829, + "scsemi_minus_pbisc_pp": 2.21155693447585, + "pbisc_wins": 101, + "scsemi_wins": 6, + "ties": 0, + "donors": 107 + }, + { + "cell_type": "Other", + "pbisc_mae_pp": 0.40468447528948653, + "scsemi_mae_pp": 0.3767538854333909, + "scsemi_minus_pbisc_pp": -0.02793058985609573, + "pbisc_wins": 36, + "scsemi_wins": 71, + "ties": 0, + "donors": 107 + } + ], + "composition_l1": { + "lora5": 0.17577899902000785, + "scsemi5": 0.2324144852627699 + }, + "overall_pbisc_win_donors": 93, + "relative_l1_reduction_percent": 24.368311716341374, + "tie_tolerance_pp": 1e-10, + "method_colors": { + "PBISC LoRA": "#D55E00", + "scSemiProfiler": "#0072B2" + }, + "unchanged_mean_composition_fractions": [ + [ + 0.6390479323723364, + 0.07479609654330255, + 0.166538380766508, + 0.09119709084308379, + 0.022818037313248645, + 0.00560246216152067 + ], + [ + 0.4526230529595016, + 0.08467912772585667, + 0.3353769470404984, + 0.07074143302180683, + 0.027246105919003092, + 0.029333333333333295 + ], + [ + 0.7067220145379024, + 0.08552377985462092, + 0.1310463136033229, + 0.04961090342679124, + 0.011994600207684288, + 0.015102388369678057 + ], + [ + 0.690181412253375, + 0.07505337487019725, + 0.11744174454828656, + 0.05893572170301138, + 0.05212731048805812, + 0.006260436137071619 + ], + [ + 0.6918943925233645, + 0.08173219106957422, + 0.12600903426791274, + 0.06422201453790234, + 0.027388785046728936, + 0.0087535825545171 + ] + ], + "raw_predictions_changed": false, + "derived_summary_changed": true, + "inference": "descriptive only; overlapping support draws are not independent training replicates" + }, + "rare_observed": { + "estimator": "median obs_pct across draws from the lora5 observed reference; no additional state filtering", + "states": [ + "C1Q-high monocyte (reference)", + "Cycling T/NK", + "Cycling activated T (CD38/HLA-DR)", + "Proliferating T (label)", + "Plasmablast (label)", + "Plasmablast (markers)", + "MS1 monocyte (IL1R2+)", + "Cycling monocyte (Ye 2026)", + "ISG-high monocyte (positive control)", + "ISG-high T", + "DN2 / ABC B", + "HSPC", + "Cytotoxic CD4 T", + "Granulocyte-like (LDG)", + "Platelet signal in monocytes (negative control)" + ], + "display_labels": [ + "C1Q-high monocyte", + "Cycling T/NK", + "Cycling activated T", + "Proliferating T (label)", + "Plasmablast (label)", + "Plasmablast (markers)", + "MS1 monocyte (IL1R2+)", + "Cycling monocyte", + "ISG-high monocyte (+ control)", + "ISG-high T", + "DN2 / ABC B", + "HSPC", + "Cytotoxic CD4 T", + "Granulocyte-like (LDG)", + "Platelet signal in mono (\u2212 control)" + ], + "cohorts": [ + "ye2026_sepsis", + "imperial_eskd", + "giroux2022", + "ji2025_chili" + ], + "state_cohort_combinations": 59, + "percentages": [ + [ + 5.329627661748478, + 4.265919811320755, + 1.167883211678832, + 4.003724394785848 + ], + [ + 0.8088838022178573, + 0.1641599699821769, + 0.7065166868085897, + 0.4877638550163989 + ], + [ + 1.4280325483644989, + 0.8014029649870511, + 1.0161920714684534, + 1.419489203113011 + ], + [ + 1.8591367139084984, + 1.0111062879164858, + 1.1166945840312674, + 1.5130984643179766 + ], + [ + 1.535064935064935, + 1.2067150881068744, + 0.296969696969697, + 0.5246887319549619 + ], + [ + 2.952380952380953, + 1.9356418619676328, + 0.98, + null + ], + [ + 13.283176060528776, + 1.526693125080015, + 0.063552589768033, + 15.595903165735567 + ], + [ + 2.232600057351194, + 1.2057783018867925, + 1.618599175985874, + 0.819366852886406 + ], + [ + 7.814485387547649, + 27.67688679245283, + 34.44962942629701, + 2.839851024208566 + ], + [ + 2.021602160216021, + 11.121443970072283, + 5.400463245154212, + 0.2429060395274373 + ], + [ + 1.911932240328892, + 2.071775312066574, + 1.740506329113924, + 1.550023485204321 + ], + [ + 0.0683982683982684, + 0.1224107465991101, + 0.0555555555555555, + 0.2747626431622187 + ], + [ + 1.1717249327698809, + 0.9929078014184396, + 0.6581009088060169, + 0.2162356969096315 + ], + [ + 0.8658008658008658, + 0.2982734857491557, + 0.4242424242424242, + 0.4869555755053184 + ], + [ + 9.212198221092756, + 2.8837536807066955, + 8.284671532846716, + 14.972067039106143 + ] + ] + }, + "rare_corrected_comparison": { + "estimator": "saved abundance_error and ranking columns; pivot only, no recomputation or imputation", + "state_cohort_combinations": 59, + "cohort_colors": { + "ye2026_sepsis": "#CC79A7", + "imperial_eskd": "#0072B2", + "giroux2022": "#009E73", + "ji2025_chili": "#E69F00" + }, + "finite_abundance_pairs": 59, + "finite_ranking_pairs": 53 + } + }, + "panel_outputs": { + "ablation": { + "png": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_ablation.png", + "pdf": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_ablation.pdf", + "svg": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_ablation.svg", + "provenance": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_ablation.provenance.json" + }, + "rank": { + "png": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_rank.png", + "pdf": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_rank.pdf", + "svg": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_rank.svg", + "provenance": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_rank.provenance.json" + }, + "behcet": { + "png": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_behcet.png", + "pdf": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_behcet.pdf", + "svg": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_behcet.svg", + "provenance": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_behcet.provenance.json" + } + }, + "method_palette": { + "schema": "pbisc.deck-method-palette.v1", + "primary": { + "PBISC": "#B2182B", + "scSemiProfiler": "#2166AC" + }, + "recolored_figures": 25, + "records": [ + { + "asset": "figures/fig1b_overview_nature.png", + "action": "recolored", + "output": "figures/method_palette_20261013/fig1b_overview_nature.png", + "proof": "figures/method_palette_20261013/fig1b_overview_nature.palette.json", + "reason": "PBISC model blocks, model arrows and optional adaptation only; preserve expression, detection, bulk and cell-type encodings.", + "preservation_pass": true + }, + { + "asset": "figures/sec02_bulk_conditioning.png", + "action": "preserved", + "reason": "Orange denotes bulk-conditioning/context, not a method comparison; preserve semantic diagram colors." + }, + { + "asset": "figures/fig1d_corpus.png", + "action": "preserved", + "reason": "Cell types and cohort composition, not method colors." + }, + { + "asset": "figures/review_style_20261013_ablation.png", + "action": "recolored", + "output": "figures/method_palette_20261013/review_style_20261013_ablation.png", + "proof": "figures/method_palette_20261013/review_style_20261013_ablation.palette.json", + "reason": "PBISC architecture and gene-panel variants; black real-R2 reference retained.", + "preservation_pass": true + }, + { + "asset": "figures/sec03_panel.png", + "action": "recolored", + "output": "figures/method_palette_20261013/sec03_panel.png", + "proof": "figures/method_palette_20261013/sec03_panel.palette.json", + "reason": "PBISC gene panels in red shades; scSemi reference line blue.", + "preservation_pass": true + }, + { + "asset": "figures/review_style_20261013_rank.png", + "action": "recolored", + "output": "figures/method_palette_20261013/review_style_20261013_rank.png", + "proof": "figures/method_palette_20261013/review_style_20261013_rank.palette.json", + "reason": "Full fine-tuning dark red and LoRA red; rank markers preserve their shapes.", + "preservation_pass": true + }, + { + "asset": "figures/fig3_healthy_pairs.png", + "action": "recolored", + "output": "figures/method_palette_20261013/fig3_healthy_pairs.png", + "proof": "figures/method_palette_20261013/fig3_healthy_pairs.palette.json", + "reason": "PBISC and its raw/calibrated inputs in red shades; scSemi blue. Neutral reference/errorbar styles retained.", + "preservation_pass": true + }, + { + "asset": "figures/fig2_healthy_pairs.png", + "action": "recolored", + "output": "figures/method_palette_20261013/fig2_healthy_pairs.png", + "proof": "figures/method_palette_20261013/fig2_healthy_pairs.palette.json", + "reason": "PBISC and its raw/calibrated inputs in red shades; scSemi blue. Neutral reference/errorbar styles retained.", + "preservation_pass": true + }, + { + "asset": "figures/sec05_condition_space.png", + "action": "preserved", + "reason": "Input-space diagnostic: measured/shift-corrected inputs, no method comparison." + }, + { + "asset": "figures/fig3_healthy_fewpair.png", + "action": "recolored", + "output": "figures/method_palette_20261013/fig3_healthy_fewpair.png", + "proof": "figures/method_palette_20261013/fig3_healthy_fewpair.palette.json", + "reason": "PBISC and its raw/calibrated inputs in red shades; scSemi blue. Neutral reference/errorbar styles retained.", + "preservation_pass": true + }, + { + "asset": "figures/review_style_20261013_healthy_composition.png", + "action": "recolored", + "output": "figures/method_palette_20261013/review_style_20261013_healthy_composition.png", + "proof": "figures/method_palette_20261013/review_style_20261013_healthy_composition.palette.json", + "reason": "Two-method paired comparison; this asset already used blue for scSemi.", + "preservation_pass": true + }, + { + "asset": "figures/fig3_healthy_copy.png", + "action": "recolored", + "output": "figures/method_palette_20261013/fig3_healthy_copy.png", + "proof": "figures/method_palette_20261013/fig3_healthy_copy.palette.json", + "reason": "PBISC and its raw/calibrated inputs in red shades; scSemi blue. Neutral reference/errorbar styles retained.", + "preservation_pass": true + }, + { + "asset": "figures/sec09_time_to_apply.png", + "action": "recolored", + "output": "figures/method_palette_20261013/sec09_time_to_apply.png", + "proof": "figures/method_palette_20261013/sec09_time_to_apply.palette.json", + "reason": "PBISC adaptation/generation red shades; scSemi training/inference blue shades.", + "preservation_pass": true + }, + { + "asset": "figures/sec09_cohorts.png", + "action": "preserved", + "reason": "Cohort metadata table, no method colors." + }, + { + "asset": "figures/fig4_disease_fewpair.png", + "action": "recolored", + "output": "figures/method_palette_20261013/fig4_disease_fewpair.png", + "proof": "figures/method_palette_20261013/fig4_disease_fewpair.palette.json", + "reason": "PBISC and its raw/calibrated inputs in red shades; scSemi blue. Neutral reference/errorbar styles retained.", + "preservation_pass": true + }, + { + "asset": "figures/fig4_disease_fivefold.png", + "action": "recolored", + "output": "figures/method_palette_20261013/fig4_disease_fivefold.png", + "proof": "figures/method_palette_20261013/fig4_disease_fivefold.palette.json", + "reason": "PBISC LoRA, zero-shot/calibrated and sc-only fine-tuning red family; scSemi blue.", + "preservation_pass": true + }, + { + "asset": "figures/sec09_cat_curves.png", + "action": "recolored", + "output": "figures/method_palette_20261013/sec09_cat_curves.png", + "proof": "figures/method_palette_20261013/sec09_cat_curves.palette.json", + "reason": "PBISC LoRA, zero-shot/calibrated and sc-only fine-tuning red family; scSemi blue.", + "preservation_pass": true + }, + { + "asset": "figures/secS_deconv_disease.png", + "action": "recolored", + "output": "figures/method_palette_20261013/secS_deconv_disease.png", + "proof": "figures/method_palette_20261013/secS_deconv_disease.palette.json", + "reason": "PBISC red, scSemi blue where present; CIBERSORTx purple and BayesPrism green.", + "preservation_pass": true + }, + { + "asset": "figures/secS_deconv_composition.png", + "action": "recolored", + "output": "figures/method_palette_20261013/secS_deconv_composition.png", + "proof": "figures/method_palette_20261013/secS_deconv_composition.palette.json", + "reason": "PBISC red, scSemi blue where present; CIBERSORTx purple and BayesPrism green.", + "preservation_pass": true + }, + { + "asset": "figures/review_style_20261013_rare_observed.png", + "action": "preserved", + "reason": "Observed-frequency heatmap encodes abundance, not a method." + }, + { + "asset": "figures/review_style_20261013_rare_corrected_comparison.png", + "action": "preserved", + "reason": "Points encode cohorts; x/y axes explicitly name methods. Preserve cohort color legend." + }, + { + "asset": "figures/review_style_20261013_behcet.png", + "action": "recolored", + "output": "figures/method_palette_20261013/review_style_20261013_behcet.png", + "proof": "figures/method_palette_20261013/review_style_20261013_behcet.palette.json", + "reason": "PBISC raw/calibrated and LoRA predictions in red family; observed native/depth-matched references black.", + "preservation_pass": true + }, + { + "asset": "figures/sec05_cells_umap.png", + "action": "preserved", + "reason": "UMAP points and legend encode cell types. Preserve images and biological labels." + }, + { + "asset": "figures/sec05_cells_metrics.png", + "action": "recolored", + "output": "figures/method_palette_20261013/sec05_cells_metrics.png", + "proof": "figures/method_palette_20261013/sec05_cells_metrics.palette.json", + "reason": "Raw-bulk and pseudobulk-input PBISC reconstructions red; observed real cells grey. Scope black replacement to four verified rectangles only.", + "preservation_pass": true + }, + { + "asset": "figures/sec09_fewpair_cells.png", + "action": "recolored", + "output": "figures/method_palette_20261013/sec09_fewpair_cells.png", + "proof": "figures/method_palette_20261013/sec09_fewpair_cells.palette.json", + "reason": "PBISC and its raw/calibrated inputs in red shades; scSemi blue. Neutral reference/errorbar styles retained.", + "preservation_pass": true + }, + { + "asset": "figures/sec09_extra_ksens.png", + "action": "recolored", + "output": "figures/method_palette_20261013/sec09_extra_ksens.png", + "proof": "figures/method_palette_20261013/sec09_extra_ksens.palette.json", + "reason": "PBISC and its raw/calibrated inputs in red shades; scSemi blue. Neutral reference/errorbar styles retained.", + "preservation_pass": true + }, + { + "asset": "figures/sec09_same_pairs.png", + "action": "recolored", + "output": "figures/method_palette_20261013/sec09_same_pairs.png", + "proof": "figures/method_palette_20261013/sec09_same_pairs.palette.json", + "reason": "PBISC and its raw/calibrated inputs in red shades; scSemi blue. Neutral reference/errorbar styles retained.", + "preservation_pass": true + }, + { + "asset": "figures/sec08_cross_cohort.png", + "action": "recolored", + "output": "figures/method_palette_20261013/sec08_cross_cohort.png", + "proof": "figures/method_palette_20261013/sec08_cross_cohort.palette.json", + "reason": "All PBISC adaptation/cohort variants in distinct red shades; existing line styles preserved.", + "preservation_pass": true + }, + { + "asset": "figures/sec08_cross_cohort_steps.png", + "action": "recolored", + "output": "figures/method_palette_20261013/sec08_cross_cohort_steps.png", + "proof": "figures/method_palette_20261013/sec08_cross_cohort_steps.palette.json", + "reason": "All PBISC adaptation/cohort variants in distinct red shades; existing line styles preserved.", + "preservation_pass": true + }, + { + "asset": "figures/sec07_no_pretraining.png", + "action": "recolored", + "output": "figures/method_palette_20261013/sec07_no_pretraining.png", + "proof": "figures/method_palette_20261013/sec07_no_pretraining.palette.json", + "reason": "Pretrained LoRA red, same-architecture scratch ablation dark red, zero-shot rose.", + "preservation_pass": true + }, + { + "asset": "figures/secA_scsemi_readout.png", + "action": "recolored", + "output": "figures/method_palette_20261013/secA_scsemi_readout.png", + "proof": "figures/method_palette_20261013/secA_scsemi_readout.palette.json", + "reason": "PBISC red; official scSemi blue and modified readout pale blue with original hatch.", + "preservation_pass": true + }, + { + "asset": "figures/sec04_sle_benchmark.png", + "action": "recolored", + "output": "figures/method_palette_20261013/sec04_sle_benchmark.png", + "proof": "figures/method_palette_20261013/sec04_sle_benchmark.palette.json", + "reason": "PBISC and its raw/calibrated inputs in red shades; scSemi blue. Neutral reference/errorbar styles retained.", + "preservation_pass": true + } + ] + }, + "pptx_sha256": "3cd501c1dea00c52e745df1c9d09131a5912a3746b3a8e557de1d453a0754443", + "pdf_status": "PPTX export verified: 34 pages, text bounds checked", + "pdf_sha256": "5e2bcfc2e16f419f2cac37ec55688f55171233cbe8f9faa817a431f4ffbdde00", + "originals_preserved": { + "PBISC-Diffusion_paper_summary_20260930.pdf": "eeb0d692c6c89cf8389e7b0294862ee54e955204b4a794db3545bafb64779141", + "PBISC-Diffusion_paper_summary_20260930.pptx": "51bc0f3a23856bd1d71d38b7bfdbe6f75b63e8054515b8b8a8b2106e9a1612e5", + "generate_deck.py": "005bec4082f16a095b2db207ec02fa8155226825ea2e243b406ed0a2108fa180", + "tools/style.py": "0d17caf276ae0e9b25285b8fd8f9b75c55621ad3a36382b705a366f327c89de6", + "data/slides_spec.json": "6b8fa055e20a3379ce2e50ca19dbbe285d81e4e95d2c42597e236945b8b2bc23", + "figures/fig1b_overview_nature.png": "e50ff68c0a08b0c4730c7a5abf2700cd1b994afc8ca2e69de60d33ff99f8c9e1" + } +} diff --git a/paper-releases/pbisc-diffusion/updates/20261013T022035Z-pbisc-red-scsemi-blue/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/data/results_review_20261013_spec.json b/paper-releases/pbisc-diffusion/updates/20261013T022035Z-pbisc-red-scsemi-blue/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/data/results_review_20261013_spec.json new file mode 100644 index 0000000000000000000000000000000000000000..e676cb95232caf99c74f1cbf1a325aa21e8ff56c --- /dev/null +++ b/paper-releases/pbisc-diffusion/updates/20261013T022035Z-pbisc-red-scsemi-blue/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/data/results_review_20261013_spec.json @@ -0,0 +1,1280 @@ +{ + "deck": { + "output_pptx": "PBISC-Diffusion_results_review_20261013.pptx", + "style_source": "generate_deck.py + tools/style.py" + }, + "slides": [ + { + "kind": "title", + "title": "PBISC-Diffusion", + "subtitle": "Results review \u00b7 existing evidence and protocol updates \u00b7 13 October 2026" + }, + { + "kind": "agenda", + "title": "Six figures: evidence available now", + "items": [ + "Fig. 1 Pretraining, diffusion and bulk conditioning", + "Fig. 2 Model design and paired-data requirements", + "Fig. 3 Healthy reconstruction: quality, composition and time", + "Fig. 4 Disease signals and deconvolution benchmarks", + "Fig. 5 Rare cell states: corrected analysis of existing outputs", + "Fig. 6 Bulk-only reproduction: awaiting the new dataset" + ] + }, + { + "kind": "content", + "section": "plan", + "section_label": "Fig. 1a", + "text_source": { + "file": "data/figure_plan_slide_text.json", + "index": 8 + }, + "title": "PBISC-Diffusion: pretraining to generated cells", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/fig1b_overview_nature.png" + ], + "keywords": [], + "footnote": "Schematic. Conditioning recipe size and the number of generated cells are distinct; no cell-type quotas.", + "sources": "tools/figure_plan.py; manuscript figures/fig01_architecture_nature.svg", + "footnote_full": "Schematic. Conditioning recipe size and the number of generated cells are distinct; no cell-type quotas.\nOriginal context: Candidate replacement for the engineering diagram (kept in this section as details and for Methods). Source: paper-release-20260928/manuscript/figures/fig01_architecture_nature.svg (scripts/build_figure1_architecture_nature.py).", + "numbers": [], + "original_slide": 7, + "figures_before_palette": [ + "figures/fig1b_overview_nature.png" + ] + }, + { + "kind": "content", + "section": "1", + "section_label": "Fig. 1b", + "title": "Bulk conditioning", + "layout": "diagram", + "figures": [ + "figures/sec02_bulk_conditioning.png" + ], + "keywords": [ + "One shared bulk condition; independent cell draws", + "Expression, depth and measured-gene masks", + "No cell-type or disease labels as model input" + ], + "footnote": "Final 6k model. Recipe size K conditions the bulk; the number of generated cells is chosen separately.", + "sources": "tools/sec02_backbone.py; data/sec02_bulk_conditioning.csv, data/sec02_slide_text.csv", + "footnote_full": "Final 6k model. Recipe size K conditions the bulk; the number of generated cells is chosen separately.\nOriginal context: Final model f01 6k 1x, part 16. Bar heights schematic.", + "original_slide": 8, + "figures_before_palette": [ + "figures/sec02_bulk_conditioning.png" + ] + }, + { + "kind": "content", + "section": "plan", + "section_label": "Fig. 1c", + "text_source": { + "file": "data/figure_plan_slide_text.json", + "index": 9 + }, + "title": "Pretraining corpus", + "layout": "full", + "figures": [ + "figures/fig1d_corpus.png" + ], + "keywords": [ + "41.4 M cells, 22 datasets, 3,199 donors", + "88.5% healthy donors; 23.6% cultured in vitro", + "Labels describe the corpus; never model input" + ], + "footnote": "Corpus labels describe the data; they are not model inputs. Dataset composition is not sampler exposure.", + "sources": "tools/figure_plan.py; manuscript figures/fig01_corpus_nature.png, source_data/corpus/", + "footnote_full": "Corpus labels describe the data; they are not model inputs. Dataset composition is not sampler exposure.\nOriginal context: Source: paper-release-20260928/manuscript/source_data/corpus/ (build_corpus_tables.py) and figures/fig01_corpus_nature.png (build_figure1_corpus.py).", + "numbers": [ + { + "value": "41,395,184 cells; 22 datasets; 3,199 donors; healthy 88.5%; in vitro 23.6%", + "source": "manuscript/source_data/corpus/corpus_summary.json" + } + ], + "original_slide": 9, + "figures_before_palette": [ + "figures/fig1d_corpus.png" + ] + }, + { + "kind": "content", + "title": "Native-pseudobulk model comparisons", + "section_label": "Fig. 2a", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/review_style_20261013_ablation.png" + ], + "keywords": [], + "footnote": "Completed held-out rescoring. Architecture: 1% detection floor; gene-panel comparison: 0.1% floor.", + "footnote_full": "Completed held-out rescoring. Architecture: 1% detection floor; gene-panel comparison: 0.1% floor.", + "sources": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_ablation.pdf", + "figures_before_palette": [ + "figures/review_style_20261013_ablation.png" + ] + }, + { + "kind": "content", + "section": "1", + "section_label": "Fig. 2b", + "title": "Gene panels: performance and coverage", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/sec03_panel.png" + ], + "keywords": [ + "6k \u2248 3k: 3k r, DEG inside 6k seed range", + "2k best L1; 15k worst r, L1", + "6k kept: first panel with CLEC9A (cDC1)" + ], + "footnote": "Existing panel benchmark. 6k: five sampling seeds; other panels: one seed. Different axis from the architecture sweep.", + "say": "Runs f01_{2k,3k,6k,p15k}_1x, part 16, Bernoulli, EMA, DDIM 50. Different gene axis from the previous slide, so the two slides cannot be compared. Loss curves are divided by each panel's own part-16 value.", + "sources": "tools/sec03_panel.py; data/sec03_panel_bench.csv, data/sec03_panel_seeds.csv, data/sec03_panel_loss.csv, data/sec03_panel_headline.csv", + "original_slide": 12, + "footnote_full": "Existing panel benchmark. 6k: five sampling seeds; other panels: one seed. Different axis from the architecture sweep.\nOriginal context: Existing panel benchmark. 6k: five sampling seeds; other panels: one seed. Different axis from the architecture sweep.", + "figures_before_palette": [ + "figures/sec03_panel.png" + ] + }, + { + "kind": "content", + "title": "Healthy bulk: full fine-tuning and LoRA rank", + "section_label": "Fig. 2c", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/review_style_20261013_rank.png" + ], + "keywords": [], + "footnote": "Original donor-level five-fold comparison. LoRA rank 16 is a quality/parameter compromise.", + "footnote_full": "Original donor-level five-fold comparison. LoRA rank 16 is a quality/parameter compromise.", + "sources": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_rank.pdf", + "figures_before_palette": [ + "figures/review_style_20261013_rank.png" + ] + }, + { + "kind": "content", + "section": "3", + "section_label": "Fig. 2d", + "text_source": { + "file": "data/fig23_healthy_slide_text.json", + "index": 6 + }, + "title": "Historical paired-data learning curve", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/fig3_healthy_pairs.png" + ], + "keywords": [ + "Coverage at every n: LoRA 0.76\u20130.84, scSemiProfiler 0.04", + "Composition L1 lower for LoRA at every n", + "5-fold: LoRA 0.16 vs 0.22 L1" + ], + "footnote": "LoRA: n training + 7 validation. scSemi: n + 7 representatives, with CV-HVG reuse. Not the strict total-n protocol.", + "sources": "tools/fig23_healthy.py; data/fig23_healthy.csv", + "refs": [ + "Wang 2024, Nat Commun" + ], + "refs_full": [ + "Wang et al. scSemiProfiler: advancing large-scale single-cell studies through semi-profiling with deep generative models and active learning. Nat Commun 15:5989 (2024). doi:10.1038/s41467-024-50150-1" + ], + "say": "Giving scSemiProfiler exactly LoRA's pairs, from five up to every training donor, does not close the gap in coverage or composition.", + "footnote_full": "LoRA: n training + 7 validation. scSemi: n + 7 representatives, with CV-HVG reuse. Not the strict total-n protocol.\nOriginal context: ImMicro, donor-level 5-fold CV pooled at each pair count: for every fold, scSemiProfiler's representatives are exactly the n pairs LoRA trained on (plus the inner-validation pairs); 'all' (open symbols, right of the dotted line) = every training donor of the fold, i.e. the 5-fold result, reported separately from the 5-pair comparison. Source: data/sec08_pairs_realism.csv, data/sec08_pairs_signal.csv, data/sec09_same_pairs.csv.", + "numbers": [ + { + "value": "pairs 5 comp_l1 scsemi 0.2184; pairs 10 comp_l1 scsemi 0.2269; pairs 20 comp_l1 scsemi 0.2232; pairs 40 comp_l1 scsemi 0.2213; pairs all comp_l1 scsemi 0.217; pairs 5 comp_l1 lora 0.1715; pairs 10 comp_l1 lora 0.1879; pairs 20 comp_l1 lora 0.1723; pairs 40 comp_l1 lora 0.1714; pairs all comp_l1 lora 0.1617; pairs 5 coverage scsemi 0.03819; pairs 10 coverage scsemi 0.03801; pairs 20 coverage scsemi 0.0372; pairs 40 coverage scsemi 0.03844; pairs all coverage scsemi 0.03844; pairs 5 coverage lora 0.8392; pairs 10 coverage lora 0.7857; pairs 20 coverage lora 0.7552; pairs 40 coverage lora 0.7784; pairs all coverage lora 0.7843; pairs 5 nk_fraction_r scsemi 0.2651; pairs 10 nk_fraction_r scsemi 0.33; pairs 20 nk_fraction_r scsemi 0.4762; pairs 40 nk_fraction_r scsemi 0.4937; pairs all nk_fraction_r scsemi 0.4245; pairs 5 nk_fraction_r lora 0.1993; pairs 10 nk_fraction_r lora 0.3972; pairs 20 nk_fraction_r lora 0.5644; pairs 40 nk_fraction_r lora 0.6993; pairs all nk_fraction_r lora 0.7019", + "source": "data/fig23_healthy.csv" + } + ], + "original_slide": 31, + "figures_before_palette": [ + "figures/fig3_healthy_pairs.png" + ] + }, + { + "kind": "content", + "section": "2", + "section_label": "Fig. 2e", + "text_source": { + "file": "data/fig23_healthy_slide_text.json", + "index": 3 + }, + "title": "Additional pairs improve donor-level information", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/fig2_healthy_pairs.png" + ], + "keywords": [ + "5 pairs: coverage 0.84 (zero-shot 0.19)", + "Donor NK fraction r 0.20 \u2192 0.70 with more pairs", + "5 pairs stop early, near 3,000 steps" + ], + "footnote": "Historical n + 7 validation design. The fixed-3000-step strict curve is a separate protocol update.", + "sources": "tools/fig23_healthy.py; data/fig23_healthy.csv", + "refs": [ + "Kynk\u00e4\u00e4nniemi 2019, NeurIPS", + "Naeem 2020, ICML" + ], + "refs_full": [ + "Kynk\u00e4\u00e4nniemi et al. Improved precision and recall metric for assessing generative models. NeurIPS 2019", + "Naeem et al. Reliable fidelity and diversity metrics for generative models. ICML 2020" + ], + "say": "Five pairs already give most of the coverage; more pairs mainly improve donor-level composition, such as the NK fraction.", + "footnote_full": "Historical n + 7 validation design. The fixed-3000-step strict curve is a separate protocol update.\nOriginal context: ImMicro, donor-level 5-fold CV, pooled scoring (107 donors); x = LoRA training pairs per fold (the last point is every training donor of the fold, 78\u201379). Calibration fitted on exactly the pairs the LoRA point saw. Selected step = early-stopping step on 7 inner-validation pairs (cap 20,000). Values: coverage LoRA 0.84 / 0.79 / 0.76 / 0.78 / 0.78; NK fraction r LoRA 0.20 / 0.40 / 0.56 / 0.70 / 0.70 at 5 / 10 / 20 / 40 / all pairs. Source: data/sec08_pairs_realism.csv, data/sec08_pairs_signal.csv, data/sec08_selected_steps_per_fold.csv.", + "numbers": [ + { + "value": "pairs 0 comp_l1 zero_shot 0.4549; pairs 5 comp_l1 zs_cal_budget 0.1916; pairs 10 comp_l1 zs_cal_budget 0.1996; pairs 20 comp_l1 zs_cal_budget 0.1911; pairs 40 comp_l1 zs_cal_budget 0.196; pairs all comp_l1 zs_cal_budget 0.1985; pairs 5 comp_l1 lora 0.1715; pairs 10 comp_l1 lora 0.1879; pairs 20 comp_l1 lora 0.1723; pairs 40 comp_l1 lora 0.1714; pairs all comp_l1 lora 0.1617; pairs 0 coverage zero_shot 0.1918; pairs 5 coverage zs_cal_budget 0.527; pairs 10 coverage zs_cal_budget 0.5287; pairs 20 coverage zs_cal_budget 0.5287; pairs 40 coverage zs_cal_budget 0.5224; pairs all coverage zs_cal_budget 0.5272; pairs 5 coverage lora 0.8392; pairs 10 coverage lora 0.7857; pairs 20 coverage lora 0.7552; pairs 40 coverage lora 0.7784; pairs all coverage lora 0.7843; pairs 0 nk_fraction_r zero_shot 0.4939; pairs 5 nk_fraction_r zs_cal_budget 0.5019; pairs 10 nk_fraction_r zs_cal_budget 0.515; pairs 20 nk_fraction_r zs_cal_budget 0.5251; pairs 40 nk_fraction_r zs_cal_budget 0.5; pairs all nk_fraction_r zs_cal_budget 0.5004; pairs 5 nk_fraction_r lora 0.1993; pairs 10 nk_fraction_r lora 0.3972; pairs 20 nk_fraction_r lora 0.5644; pairs 40 nk_fraction_r lora 0.6993; pairs all nk_fraction_r lora 0.7019; pairs 5 selected_step_median_k lora 3.25; pairs 10 selected_step_median_k lora 5.75; pairs 20 selected_step_median_k lora 19.5; pairs 40 selected_step_median_k lora 20; pairs all selected_step_median_k lora 20", + "source": "data/fig23_healthy.csv" + } + ], + "original_slide": 26, + "figures_before_palette": [ + "figures/fig2_healthy_pairs.png" + ] + }, + { + "kind": "content", + "section": "2", + "section_label": "Fig. 3a", + "title": "Measured bulk differs from pseudobulk", + "layout": "figure_keywords", + "figures": [ + "figures/sec05_condition_space.png" + ], + "keywords": [ + "Real bulk: 2.46 D from own pseudobulk", + "95% of squared offset is one shift", + "Shift removed (diagnostic): own donor 7.5% \u2192 75.7%" + ], + "footnote": "Existing paired-input diagnostic; mean-shift removal uses paired pseudobulks and is not a deployment method.", + "say": "The two pseudobulk clusters on PC1 are sc pools DB1-3 vs DB4-7, so the pairing lines cross mainly because of sc batch. Shift removal subtracts the mean offset of the other 106 donors; a bulk-only deployment would not have them.", + "sources": "tools/sec05_condition_space.py; data/sec05_condition_space_numbers.csv, data/sec05_condition_space_coords.csv", + "footnote_full": "Existing paired-input diagnostic; mean-shift removal uses paired pseudobulks and is not a deployment method.\nOriginal context: ImMicro, 107 healthy donors. Model inputs (log1p CPM, 5,508 genes), not learned ctx. D = median donor-to-donor distance; PCA-30 values are lower bounds (full space 3.19 D, 0.91 shared). Shift removal is a diagnostic that uses paired pseudobulks. Pseudobulk PC1 is sc pool batch (pool R\u00b2 0.96 on PC1).", + "original_slide": 17, + "figures_before_palette": [ + "figures/sec05_condition_space.png" + ] + }, + { + "kind": "content", + "section": "3", + "section_label": "Fig. 3b", + "text_source": { + "file": "data/fig23_healthy_slide_text.json", + "index": 5 + }, + "title": "Healthy donors: exactly five paired samples", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/fig3_healthy_fewpair.png" + ], + "keywords": [ + "LoRA-5 ahead on 6 of 6 metrics", + "Coverage 0.79 vs 0.04", + "E_N 1.5 vs 3.6 (scSemiProfiler's own metric)" + ], + "footnote": "ImMicro: exactly 5 pairs, no extra validation, 3000-step LoRA; five draws, remaining 102 donors per draw.", + "sources": "tools/fig23_healthy.py; data/fig23_healthy.csv", + "refs": [ + "Wang 2024, Nat Commun", + "Lopez-Paz 2017, ICLR" + ], + "refs_full": [ + "Wang et al. scSemiProfiler: advancing large-scale single-cell studies through semi-profiling with deep generative models and active learning. Nat Commun 15:5989 (2024). doi:10.1038/s41467-024-50150-1", + "Lopez-Paz & Oquab. Revisiting classifier two-sample tests. ICLR 2017" + ], + "say": "With the same five healthy donors, LoRA-5 beats scSemiProfiler on all six cell-level metrics, including scSemiProfiler's own error metric.", + "footnote_full": "ImMicro: exactly 5 pairs, no extra validation, 3000-step LoRA; five draws, remaining 102 donors per draw.\nOriginal context: Main comparison on healthy donors: per draw, LoRA-5 and scSemiProfiler get the same 5 paired donors (scSemiProfiler: as its representatives) and are tested on the other 102 from their real bulk; 5 draws. Zero-shot (no pairs) shown for reference. Dashed: real vs real cells. Values (LoRA-5 / scSemiProfiler-5): coverage 0.792 / 0.036; precision 0.952 / 0.518; comp_l1 0.176 / 0.233; c2st_knn_auc 0.943 / 0.999; mmd2_bw1 0.023 / 0.186; en_all_sq 1.547 / 3.649. Source: data/sec09_fewpair_cells.csv, data/sec09_extra_fewpair.csv.", + "numbers": [ + { + "value": "5 donors (mean of 5 draws) coverage zs_raw 0.19; 5 donors (mean of 5 draws) coverage scsemi 0.03583; 5 donors (mean of 5 draws) coverage lora 0.7925; 5 donors (mean of 5 draws) precision zs_raw 0.6772; 5 donors (mean of 5 draws) precision scsemi 0.518; 5 donors (mean of 5 draws) precision lora 0.9521; 5 donors (mean of 5 draws) comp_l1 zs_raw 0.4451; 5 donors (mean of 5 draws) comp_l1 scsemi 0.2325; 5 donors (mean of 5 draws) comp_l1 lora 0.1756; 5 donors (mean of 5 draws) c2st_knn_auc zs_raw 0.9966; 5 donors (mean of 5 draws) c2st_knn_auc scsemi 0.9988; 5 donors (mean of 5 draws) c2st_knn_auc lora 0.9427; 5 donors (mean of 5 draws) mmd2_bw1 zs_raw 0.1026; 5 donors (mean of 5 draws) mmd2_bw1 scsemi 0.1865; 5 donors (mean of 5 draws) mmd2_bw1 lora 0.02308; 5 donors (mean of 5 draws) en_all_sq zs_raw 7.236; 5 donors (mean of 5 draws) en_all_sq scsemi 3.649; 5 donors (mean of 5 draws) en_all_sq lora 1.547", + "source": "data/fig23_healthy.csv" + } + ], + "original_slide": 30, + "figures_before_palette": [ + "figures/fig3_healthy_fewpair.png" + ] + }, + { + "kind": "content", + "title": "Composition: error and donor-level wins", + "section_label": "Fig. 3c", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/review_style_20261013_healthy_composition.png" + ], + "keywords": [], + "footnote": "Exactly five pairs per draw. Absolute errors averaged within each held-out donor; 107 donors, equal weight. Descriptive.", + "footnote_full": "Exactly five pairs per draw. Absolute errors averaged within each held-out donor; 107 donors, equal weight. Descriptive.", + "sources": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_healthy_composition.pdf", + "figures_before_palette": [ + "figures/review_style_20261013_healthy_composition.png" + ] + }, + { + "kind": "content", + "section": "3", + "section_label": "Fig. 3d", + "text_source": { + "file": "data/fig23_healthy_slide_text.json", + "index": 7 + }, + "title": "Representative-dependent composition", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/fig3_healthy_copy.png" + ], + "keywords": [ + "Composition varies with the selected representative", + "Compare generated fractions with held-out observed fractions" + ], + "footnote": "Existing healthy five-draw diagnostic. This evaluates composition dependence, not literal copying of every cell.", + "sources": "tools/fig23_healthy.py; data/fig23_healthy.csv", + "refs": [ + "Wang 2024, Nat Commun" + ], + "refs_full": [ + "Wang et al. scSemiProfiler: advancing large-scale single-cell studies through semi-profiling with deep generative models and active learning. Nat Commun 15:5989 (2024). doi:10.1038/s41467-024-50150-1" + ], + "say": "scSemiProfiler does worse than simply copying its representative's composition, and its error follows that representative; LoRA does better than the copy.", + "footnote_full": "Existing healthy five-draw diagnostic. This evaluates composition dependence, not literal copying of every cell.\nOriginal context: ImMicro, 5 donors. scSemiProfiler assigns each test sample one representative; 'copy' uses that representative's observed composition as the prediction. Right: Spearman over test samples between each method's composition error and the copy's error. Source: data/sec09_scsemi_representative.csv (scsemi/analysis/representative_dependence_20261002).", + "numbers": [ + { + "value": "5 donors (mean of 5 draws) comp_l1 copy 0.1871; 5 donors (mean of 5 draws) comp_l1 scsemi 0.2325; 5 donors (mean of 5 draws) comp_l1 lora 0.1756; spearman_with_copy_error scsemi 0.5736; spearman_with_copy_error lora 0.3509", + "source": "data/fig23_healthy.csv" + } + ], + "original_slide": 32, + "figures_before_palette": [ + "figures/fig3_healthy_copy.png" + ] + }, + { + "kind": "content", + "section": "3", + "section_label": "Fig. 3e", + "text_source": { + "file": "data/sec09_time_to_apply_slide_text.json", + "index": 0 + }, + "title": "Adaptation and generation time", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/sec09_time_to_apply.png" + ], + "keywords": [ + "LoRA-5: 3\u201311 min per draw, every test sample", + "scSemiProfiler-5: 57 min\u20134.1 h per draw", + "17\u201359\u00d7 less GPU time, same 5 donors" + ], + "footnote": "Existing measurements: adaptation + target generation; pretraining, GPU wait and scoring excluded.", + "sources": "tools/sec09_time_to_apply.py; data/sec09_time_to_apply.csv, data/sec09_time_to_apply_runs.csv; data/sec06_ft_method.csv (pretrain_wall_hours)", + "footnote_full": "Existing measurements: adaptation + target generation; pretraining, GPU wait and scoring excluded.\nOriginal context: Same 25 draws as the '5 donors' slides (5 cohorts x 5 draws; same 5 training donors and test samples for both methods). GPU time on one GB10, every step alone on the GPU (shared lock). LoRA-5: fine-tune = RUN.json wall_seconds of the 3,000-step LoRA; generate = wall clock of the sampling job in the queue log (model load + DDIM 50, 1,500 cells per test sample). scSemiProfiler-5 (upstream c2aea02, default budget): stage 1 + stage 2 pretraining of every representative (model receipts) + its per-target inference (5 ministages x 150 epochs, export receipt total_target_seconds). Not counted on either side: CPU data preparation, scoring, work lost to interruptions, and PBISC's one-time pretraining (15.9 h; 'Fine-tuning with paired bulk + sc' slide), which every cohort reuses. scSemiProfiler time grows with the number of test samples because it trains one model per sample. Per-draw ratio scSemiProfiler / LoRA: ImMicro 15\u201319\u00d7; Sepsis 26\u201347\u00d7; Imperial 19\u201326\u00d7; Giroux 54\u201363\u00d7; Ji 11\u201320\u00d7 (all 25 draws 11\u201363\u00d7; keyword and figure = mean of each cohort's 5 draws).", + "numbers": [ + { + "value": "ImMicro: LoRA-5 8.4 min (fine-tune 1.9 + generate 6.5), scSemiProfiler-5 141 min (representatives 15 + per-sample models 126); 15\u201319\u00d7 per draw; per test sample 4.9 s vs 83 s", + "source": "data/sec09_time_to_apply.csv" + }, + { + "value": "Sepsis: LoRA-5 6.7 min (fine-tune 1.8 + generate 4.9), scSemiProfiler-5 244 min (representatives 30 + per-sample models 214); 26\u201347\u00d7 per draw; per test sample 5.2 s vs 190 s", + "source": "data/sec09_time_to_apply.csv" + }, + { + "value": "Imperial: LoRA-5 10.8 min (fine-tune 1.8 + generate 9.0), scSemiProfiler-5 229 min (representatives 32 + per-sample models 197); 19\u201326\u00d7 per draw; per test sample 4.5 s vs 96 s", + "source": "data/sec09_time_to_apply.csv" + }, + { + "value": "Giroux: LoRA-5 2.6 min (fine-tune 1.9 + generate 0.7), scSemiProfiler-5 156 min (representatives 117 + per-sample models 38); 54\u201363\u00d7 per draw; per test sample 14.8 s vs 872 s", + "source": "data/sec09_time_to_apply.csv" + }, + { + "value": "Ji: LoRA-5 3.4 min (fine-tune 1.8 + generate 1.6), scSemiProfiler-5 57 min (representatives 24 + per-sample models 33); 11\u201320\u00d7 per draw; per test sample 8.1 s vs 137 s", + "source": "data/sec09_time_to_apply.csv" + }, + { + "value": "PBISC one-time pretraining 15.9 h (not counted per draw)", + "source": "data/sec06_ft_method.csv (pretrain_wall_hours)" + } + ], + "refs": [ + "Wang 2024, Nat Commun" + ], + "refs_full": [ + "Wang et al. scSemiProfiler: advancing large-scale single-cell studies through semi-profiling with deep generative models and active learning. Nat Commun 15:5989 (2024). doi:10.1038/s41467-024-50150-1" + ], + "original_slide": 34, + "figures_before_palette": [ + "figures/sec09_time_to_apply.png" + ] + }, + { + "kind": "content", + "section": "4", + "section_label": "Fig. 4a", + "sec09_index": 0, + "title": "Paired cohorts and evaluation units", + "layout": "full", + "figures": [ + "figures/sec09_cohorts.png" + ], + "keywords": [ + "403 paired samples, 275 donors", + "Identity screening where available; donor-level splits", + "Ji bulk: whole blood, not PBMC" + ], + "footnote": "Existing paired-cohort map. Five donors means 5 pairs in Sepsis, 12-15 Imperial, 12-14 Giroux, 7-9 Ji.", + "footnote_full": "Existing paired-cohort map. Five donors means 5 pairs in Sepsis, 12-15 Imperial, 12-14 Giroux, 7-9 Ji.\nOriginal context: Usable pair = passed the pairing identity check and >= 200 cells (fold >= 0 in data//folds.csv); dropped 15 identity-fail, 11 borderline and 1 low-cell pairs of 430. ImMicro: no identity flag, all used. Imperial, Giroux, Ji: several pairs per donor. Case vs control = samples in scores//pooled/summary.json. Giroux 'moderate' = GEO 'severe' (paper term); its moderate vs mild subjects are 4F/0M vs 0F/5M (scored), so its disease metrics exclude sex-linked genes. Labels: data//QC_REPORT.md; ImMicro: GEO GSE314922/GSE314416.", + "sources": "tools/sec09_paired_validation.py; data/sec09_cohorts.csv, data/sec09_quoted_numbers.csv, data/sec09_slide_text.json", + "numbers": [ + { + "value": "403 paired samples (sum of usable rows)", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data//fewshot_n5/DRAWS.json (n_usable_rows)" + }, + { + "value": "275 donors (sum of usable donors)", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data//fewshot_n5/DRAWS.json (n_usable_donors)" + }, + { + "value": "dropped 15 identity-fail, 11 borderline, 1 < 200 cells, of 430", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data//features.npz (identity_pass, identity_check); samples.csv (n_cells); /home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/tools/prepare_dataset.py (MIN_CELLS)" + }, + { + "value": "ImMicro 107 pairs / 107 donors", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data/immicro/fewshot_n5/DRAWS.json" + }, + { + "value": "Sepsis 82 pairs / 82 donors; sepsis 61 vs control 21", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data/ye2026_sepsis/fewshot_n5/DRAWS.json; /home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores/ye2026_sepsis/pooled/summary.json" + }, + { + "value": "Imperial 157 pairs / 57 donors; COVID+ 113 vs COVID\u2212 34 (+10 recovery, neither)", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data/imperial_eskd/fewshot_n5/DRAWS.json; /home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores/imperial_eskd/pooled/summary.json" + }, + { + "value": "Giroux 24 pairs / 9 donors; moderate 11 vs mild 13", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data/giroux2022/fewshot_n5/DRAWS.json; /home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores/giroux2022/pooled/summary.json" + }, + { + "value": "Ji 33 pairs / 20 donors; ChILI 18 vs ICI control 15", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data/ji2025_chili/fewshot_n5/DRAWS.json; /home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores/ji2025_chili/pooled/summary.json" + } + ], + "original_slide": 37, + "figures_before_palette": [ + "figures/sec09_cohorts.png" + ] + }, + { + "kind": "content", + "section": "4", + "section_label": "Fig. 4b", + "text_source": { + "file": "data/fig4_disease_slide_text.json", + "index": 0 + }, + "title": "Disease signals: five-donor draws", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/fig4_disease_fewpair.png" + ], + "keywords": [ + "vs scSemiProfiler-5: logFC r 13/20, partial r 14/20 draws", + "Beyond bulk: partial r 0.60 vs 0.43 (Imperial)", + "Above zero-shot and calibration in 4/4 cohorts" + ], + "footnote": "Existing five-donor protocol; sample counts differ by cohort. Original sample-level estimator; corrected update separate.", + "sources": "tools/fig4_disease.py; data/fig4_disease.csv", + "refs": [ + "Wang 2024, Nat Commun", + "Vi\u00f1as Torn\u00e9 2026, Nat Biotechnol" + ], + "refs_full": [ + "Wang et al. scSemiProfiler: advancing large-scale single-cell studies through semi-profiling with deep generative models and active learning. Nat Commun 15:5989 (2024). doi:10.1038/s41467-024-50150-1", + "Vi\u00f1as Torn\u00e9 et al. Systema: a framework for evaluating genetic perturbation response prediction beyond systematic variation. Nat Biotechnol 44:1050-1059 (2026). doi:10.1038/s41587-025-02777-8" + ], + "say": "The main comparison uses five pairs, the situation a lab is in. With the same five donors, LoRA-5 is above both zero-shot versions in every cohort and ahead of scSemiProfiler in most draws. The bold panel is partial r: the cell-type signal the bulk itself does not carry, which is what single cells add.", + "footnote_full": "Existing five-donor protocol; sample counts differ by cohort. Original sample-level estimator; corrected update separate.\nOriginal context: Main comparison: 5 donors per draw, LoRA\u20135 vs scSemiProfiler with the same 5 donors as representatives (and the two zero\u2013shot PBISC versions), 5 draws per cohort, every other usable sample tested. logFC r = Pearson r of case\u2013vs\u2013control logFC (mean CP10K) with the observed cells, mean over T, B, monocyte and NK. Partial r = the same after both logFCs are residualised on the input bulk's logFC: the cell\u2013type signal the bulk does not carry. Top\u201350 DEG: LoRA\u20135 ahead in 12/20 draws, tied in 8 (Giroux and Ji: few DEG in the observed cells, mostly 0 for every method). Giroux's 5\u2013donor test sets hold 10\u201312 samples from 4 subjects (2 vs 2), so its disease metrics are noisy; there scSemiProfiler\u20135 is ahead. Giroux: sex\u2013linked genes excluded (severity is confounded with sex). Per cohort, mean over draws (r / partial r / DEG): Sepsis LoRA\u20135 0.35 / 0.30 / 16.4, scSemiProfiler\u20135 0.40 / 0.33 / 5.2; Imperial LoRA\u20135 0.75 / 0.60 / 113.6, scSemiProfiler\u20135 0.56 / 0.43 / 46.8; Giroux LoRA\u20135 0.30 / 0.19 / 0.2, scSemiProfiler\u20135 0.40 / 0.24 / 0.0; Ji LoRA\u20135 0.33 / 0.30 / 0.2, scSemiProfiler\u20135 0.15 / 0.09 / 0.0. Source: data/fig4_disease.csv (scores[_nosex]//fewshot_n5/draw[_scsemi5]).", + "numbers": [ + { + "value": "Sepsis LoRA\u20135 0.35 / 0.30 / 16.4, scSemiProfiler\u20135 0.40 / 0.33 / 5.2; Imperial LoRA\u20135 0.75 / 0.60 / 113.6, scSemiProfiler\u20135 0.56 / 0.43 / 46.8; Giroux LoRA\u20135 0.30 / 0.19 / 0.2, scSemiProfiler\u20135 0.40 / 0.24 / 0.0; Ji LoRA\u20135 0.33 / 0.30 / 0.2, scSemiProfiler\u20135 0.15 / 0.09 / 0.0", + "source": "data/fig4_disease.csv" + } + ], + "original_slide": 38, + "figures_before_palette": [ + "figures/fig4_disease_fewpair.png" + ] + }, + { + "kind": "content", + "section": "4", + "section_label": "Fig. 4c", + "text_source": { + "file": "data/fig4_disease_slide_text.json", + "index": 1 + }, + "title": "Disease signals: five-fold validation", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/fig4_disease_fivefold.png" + ], + "keywords": [ + "logFC r up to 0.82 (Imperial)", + "vs scSemiProfiler: r 3/4, partial r 3/4, DEG 4/4 cohorts", + "Top PBISC version on logFC r in 4/4 cohorts" + ], + "footnote": "Original donor-level split and original signal estimator. These results are distinct from the exact-five update.", + "sources": "tools/fig4_disease.py; data/fig4_disease.csv", + "refs": [ + "Wang 2024, Nat Commun", + "Vi\u00f1as Torn\u00e9 2026, Nat Biotechnol" + ], + "refs_full": [ + "Wang et al. scSemiProfiler: advancing large-scale single-cell studies through semi-profiling with deep generative models and active learning. Nat Commun 15:5989 (2024). doi:10.1038/s41467-024-50150-1", + "Vi\u00f1as Torn\u00e9 et al. Systema: a framework for evaluating genetic perturbation response prediction beyond systematic variation. Nat Biotechnol 44:1050-1059 (2026). doi:10.1038/s41587-025-02777-8" + ], + "say": "With every fold's pairs, LoRA reaches a disease logFC r of 0.82 in Imperial and is the best PBISC version on logFC r in every cohort. This is reported separately from the five-pair comparison.", + "footnote_full": "Original donor-level split and original signal estimator. These results are distinct from the exact-five update.\nOriginal context: Reported separately from the 5-pair comparison: donor-level 5-fold CV, each sample generated by the fold model that never trained on it, one pooled scoring call per method; scSemiProfiler uses exactly LoRA's training pairs as representatives. Per cohort LoRA vs scSemiProfiler (r / partial r / DEG): Sepsis 0.53 / 0.47 / 23 vs 0.44 / 0.36 / 6; Imperial 0.82 / 0.69 / 143 vs 0.71 / 0.54 / 87; Giroux 0.46 / 0.32 / 5 vs 0.54 / 0.39 / 0; Ji 0.48 / 0.45 / 15 vs 0.27 / 0.23 / 0. Giroux: sex-linked genes excluded. Source: data/fig4_disease.csv (scores[_nosex]//pooled, scsemi_cv_20261002/pooled; sepsis DEG from pooled_deg).", + "numbers": [ + { + "value": "Sepsis 0.53 / 0.47 / 23 vs 0.44 / 0.36 / 6; Imperial 0.82 / 0.69 / 143 vs 0.71 / 0.54 / 87; Giroux 0.46 / 0.32 / 5 vs 0.54 / 0.39 / 0; Ji 0.48 / 0.45 / 15 vs 0.27 / 0.23 / 0", + "source": "data/fig4_disease.csv" + } + ], + "original_slide": 39, + "figures_before_palette": [ + "figures/fig4_disease_fivefold.png" + ] + }, + { + "kind": "content", + "section": "4", + "section_label": "Fig. 4d", + "text_source": { + "file": "data/sec09_cat_curves_slide_text.json", + "index": 0 + }, + "title": "Disease-gene agreement across list lengths", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/sec09_cat_curves.png" + ], + "keywords": [ + "LoRA top model: 3/4 (5 donors), 2/4 (5-fold)", + "LoRA-5 above scSemiProfiler-5: 3/4 cohorts (k = 100)", + "Chance at k = 100: 0.012\u20130.016" + ], + "footnote": "Existing CAT curves; ranking agreement and chance line. Patient-aware reanalysis reuses the same predictions.", + "sources": "tools/sec09_cat_curves.py; data/sec09_cat_curves.csv", + "refs": [ + "Irizarry 2005, Nat Methods", + "Webber 2010, ACM TOIS", + "Squair 2021, Nat Commun" + ], + "refs_full": [ + "Irizarry et al. Multiple-laboratory comparison of microarray platforms. Nat Methods 2:345-350 (2005). doi:10.1038/nmeth756", + "Webber, Moffat & Zobel. A similarity measure for indefinite rankings. ACM Trans Inf Syst 28(4):20 (2010). doi:10.1145/1852102.1852106", + "Squair et al. Confronting false discoveries in single-cell differential expression. Nat Commun 12:5692 (2021). doi:10.1038/s41467-021-25960-2" + ], + "footnote_full": "Existing CAT curves; ranking agreement and chance line. Patient-aware reanalysis reuses the same predictions.\nOriginal context: Correspondence at the top (Irizarry et al. 2005): per cell type, every gene tested on both sides is ranked by |limma-voom moderated t| (case vs control, same donor x type pseudobulks as the DEG panels); CAT(k) = share of the observed top-k genes that are also in the generated top-k with the same logFC sign. Mean over T / B / Mono / NK; 5 donors = mean over 5 draws (scSemiProfiler: its 25 draws), the main comparison; 5-fold = pooled call, shown separately. Chance = k / (2G), G = genes tested on both sides (0.012\u20130.016 at k = 100). No significance threshold, so cohorts where few genes pass FDR (Giroux, Ji) still get a curve. Rank-biased overlap (Webber et al. 2010, p = 0.98) gives the same order: scores_contrast/{fewshot,pooled}_rbo.csv. Giroux: sex-linked genes excluded. Sepsis CAT100 5 donors LoRA 0.092 / scSemi 0.037 / calibration 0.069 / raw 0.028 / chance 0.012; 5-fold LoRA 0.163; Imperial CAT100 5 donors LoRA 0.243 / scSemi 0.133 / calibration 0.198 / raw 0.179 / chance 0.012; 5-fold LoRA 0.270; Giroux CAT100 5 donors LoRA 0.048 / scSemi 0.075 / calibration 0.050 / raw 0.048 / chance 0.016; 5-fold LoRA 0.068; Ji CAT100 5 donors LoRA 0.091 / scSemi 0.030 / calibration 0.021 / raw 0.021 / chance 0.014; 5-fold LoRA 0.152.", + "numbers": [ + { + "value": "Sepsis CAT100 5 donors LoRA 0.092 / scSemi 0.037 / calibration 0.069 / raw 0.028 / chance 0.012; 5-fold LoRA 0.163; Imperial CAT100 5 donors LoRA 0.243 / scSemi 0.133 / calibration 0.198 / raw 0.179 / chance 0.012; 5-fold LoRA 0.270; Giroux CAT100 5 donors LoRA 0.048 / scSemi 0.075 / calibration 0.050 / raw 0.048 / chance 0.016; 5-fold LoRA 0.068; Ji CAT100 5 donors LoRA 0.091 / scSemi 0.030 / calibration 0.021 / raw 0.021 / chance 0.014; 5-fold LoRA 0.152", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores_contrast/fewshot_cat.csv; /home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores_contrast/pooled_cat.csv" + } + ], + "original_slide": 40, + "figures_before_palette": [ + "figures/sec09_cat_curves.png" + ] + }, + { + "kind": "content", + "section": "4", + "section_label": "Fig. 4e", + "text_source": { + "file": "data/secS_deconvolution_slide_text.json", + "index": 1 + }, + "title": "Cell-type expression: deconvolution comparison", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/secS_deconv_disease.png" + ], + "keywords": [ + "logFC r: PBISC ahead in 15/20 draws (BayesPrism), 17/17 (CIBERSORTx)", + "Beyond bulk: PBISC ahead in 18/20 and 17/17", + "Top-50 DEG mixed: PBISC ahead in 6/20 and 6/17" + ], + "footnote": "Completed BayesPrism/CIBERSORTx comparisons; same five-donor support, method-supported expression metrics.", + "sources": "tools/secS_deconvolution.py; data/secS_deconvolution.csv; plans/bayesprism_paired_20261006/results/{RESULTS_full.md, RESULTS_cibersortx.md, TABLE_*_full.csv, TABLE_*_cibersortx.csv}", + "refs": [ + "Chu 2022, Nat Cancer", + "Newman 2019, Nat Biotechnol" + ], + "refs_full": [ + "Chu et al. Cell type and gene expression deconvolution with BayesPrism enables Bayesian integrative analysis across bulk and single-cell RNA sequencing in oncology. Nat Cancer 3:505-517 (2022). doi:10.1038/s43018-022-00356-3", + "Newman et al. Determining cell type abundance and expression from bulk tissues with digital cytometry. Nat Biotechnol 37:773-782 (2019). doi:10.1038/s41587-019-0114-2" + ], + "say": "Both methods also estimate expression per cell type: BayesPrism's posterior and CIBERSORTx HiRes. Their case-versus-control changes per cell type agree less with the truth than PBISC's in most draws (15 of 20 for BayesPrism, all 17 scored for CIBERSORTx; BayesPrism is ahead in Giroux). Beyond what the bulk itself carries, they add little: partial r at most 0.19, against 0.19 to 0.60 for PBISC. Top-50 DEG overlap is mixed: deconvolution is ahead in sepsis, PBISC in Imperial, and Ji and Giroux are near zero for every method.", + "footnote_full": "Completed BayesPrism/CIBERSORTx comparisons; same five-donor support, method-supported expression metrics.\nOriginal context: Cell-type deconvolution, not single cells. Case-vs-control per-cell-type logFC (mean CP10K; T, B, Monocyte, NK) from BayesPrism's posterior cell-type expression Z and CIBERSORTx HiRes, scored with the benchmark's own functions; partial r = the same r after both logFCs are residualised on the input bulk's logFC (the cell-type signal the bulk does not carry); top-50 DEG = Welch, summed over types (many ties at 0 in Ji and Giroux). 5-donor draws (the 5-pair comparison), mean over draws with a score. Giroux excludes sex-linked genes, as everywhere in the deck. CIBERSORTx: official Docker images (x86, run under qemu emulation on server_1), single-cell reference mode with S-mode batch correction (up to 400 reference cells per type), fractions in cell units (divided by each type's mean UMI per reference cell). CIBERSORTx HiRes gives per-sample cell-type expression for the genes it imputes per type (others are written as 1 and count as zero here); it imputed no gene of the four contrast types in 1 Ji ChILI and 2 Giroux draws, which therefore have no CIBERSORTx disease score. Source: plans/bayesprism_paired_20261006/results/RESULTS_full.md, RESULTS_cibersortx.md.", + "numbers": [ + { + "value": "Sepsis: logFC r PBISC 0.350 / BayesPrism 0.241 / CIBERSORTx 0.069; partial r 0.305 / 0.095 / -0.010; DEG 16.4 / 27.0 / 25.4; Ji ChILI: logFC r PBISC 0.333 / BayesPrism 0.193 / CIBERSORTx 0.158; partial r 0.300 / 0.032 / 0.005; DEG 0.2 / 2.0 / 0.0; Imperial: logFC r PBISC 0.749 / BayesPrism 0.598 / CIBERSORTx 0.283; partial r 0.603 / 0.187 / 0.061; DEG 113.6 / 72.4 / 30.8; Giroux: logFC r PBISC 0.297 / BayesPrism 0.371 / CIBERSORTx 0.095; partial r 0.189 / 0.117 / 0.029; DEG 0.2 / 0.2 / 0.0", + "source": "data/secS_deconvolution.csv" + }, + { + "value": "draws where PBISC LoRA-5 beats BayesPrism: logFC r 15/20, partial r 18/20, top-50 DEG 6 (BayesPrism ahead 8, ties 6); beats CIBERSORTx: logFC r 17/17, partial r 17/17, top-50 DEG 6 (CIBERSORTx ahead 4, ties 7)", + "source": "plans/bayesprism_paired_20261006/results/TABLE_fewshot_full.csv, TABLE_fewshot_cibersortx.csv" + } + ], + "original_slide": 41, + "figures_before_palette": [ + "figures/secS_deconv_disease.png" + ] + }, + { + "kind": "content", + "section": "4", + "section_label": "Fig. 4f", + "text_source": { + "file": "data/secS_deconvolution_slide_text.json", + "index": 0 + }, + "title": "Composition: deconvolution comparison", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/secS_deconv_composition.png" + ], + "keywords": [ + "PBISC closer than BayesPrism in 22/25 draws, than CIBERSORTx in 24/25", + "Composition compared with held-out observed cells", + "Five-fold: PBISC lowest error in 4 of 5 cohorts" + ], + "footnote": "Completed deconvolution outputs already exist. Verified identical Sepsis supports use adoption instead of method reruns.", + "sources": "tools/secS_deconvolution.py; data/secS_deconvolution.csv; plans/bayesprism_paired_20261006/results/{RESULTS_full.md, RESULTS_cibersortx.md, TABLE_*_full.csv, TABLE_*_cibersortx.csv}", + "refs": [ + "Chu 2022, Nat Cancer", + "Newman 2019, Nat Biotechnol", + "Wang 2024, Nat Commun" + ], + "refs_full": [ + "Chu et al. Cell type and gene expression deconvolution with BayesPrism enables Bayesian integrative analysis across bulk and single-cell RNA sequencing in oncology. Nat Cancer 3:505-517 (2022). doi:10.1038/s43018-022-00356-3", + "Newman et al. Determining cell type abundance and expression from bulk tissues with digital cytometry. Nat Biotechnol 37:773-782 (2019). doi:10.1038/s41587-019-0114-2", + "Wang et al. scSemiProfiler: advancing large-scale single-cell studies through semi-profiling with deep generative models and active learning. Nat Commun 15:5989 (2024). doi:10.1038/s41467-024-50150-1" + ], + "say": "For reference only. Could standard bulk deconvolution do the same job with the same paired donors? BayesPrism and CIBERSORTx used the paired donors' own cells as the reference and the test samples' real bulk as the mixture, scored with the same truth and code as PBISC. PBISC estimates the cell-type mix more closely in almost every draw. On pseudobulk BayesPrism recovers the right mix, so the gap comes from real bulk, which PBISC learns from the pairs. Deconvolution also returns proportions, not cells.", + "footnote_full": "Completed deconvolution outputs already exist. Verified identical Sepsis supports use adoption instead of method reruns.\nOriginal context: Cell-type deconvolution (cell-type fractions and per-type expression, not single cells) given the same paired donors as PBISC LoRA, on the paired benchmark's own splits, observed truth and scorer: 5-donor draws (reference = the draw's 5 training donors' observed cells, mixture = the test samples' real bulk) and 5-fold CV (reference = each fold's LoRA training + inner-validation pairs; pooled). BayesPrism: recommended gene set (protein-coding minus chrX/Y/M and ribosomal genes, 15-19k), reference states = gate | AIFI L2; its better theta (initial or final) in cell units is shown. CIBERSORTx: official Docker images (x86, run under qemu emulation on server_1), single-cell reference mode with S-mode batch correction (up to 400 reference cells per type), fractions in cell units (divided by each type's mean UMI per reference cell). CIBERSORTx HiRes gives per-sample cell-type expression for the genes it imputes per type (others are written as 1 and count as zero here); it imputed no gene of the four contrast types in 1 Ji ChILI and 2 Giroux draws, which therefore have no CIBERSORTx disease score. Composition L1 over the six-type gate, lower is better. 5-fold: PBISC has the lowest L1 except in Giroux (scSemiProfiler). Pipeline check: on pseudobulk mixtures BayesPrism recovers the observed mRNA shares (e.g. T 0.39 vs 0.38), so the error comes from real bulk. Source: plans/bayesprism_paired_20261006/results/RESULTS_full.md, RESULTS_cibersortx.md.", + "numbers": [ + { + "value": "ImMicro: 5 donors L1 PBISC 0.176 / BayesPrism 0.537 / CIBERSORTx 0.491; 5-fold PBISC 0.162 / BayesPrism 0.575 / CIBERSORTx 0.507 / scSemiProfiler 0.217; Sepsis: 5 donors L1 PBISC 0.466 / BayesPrism 0.446 / CIBERSORTx 0.582; 5-fold PBISC 0.362 / BayesPrism 0.475 / CIBERSORTx 0.607 / scSemiProfiler 0.502; Ji ChILI: 5 donors L1 PBISC 0.513 / BayesPrism 0.893 / CIBERSORTx 0.827; 5-fold PBISC 0.504 / BayesPrism 0.883 / CIBERSORTx 0.914 / scSemiProfiler 0.575; Imperial: 5 donors L1 PBISC 0.503 / BayesPrism 0.845 / CIBERSORTx 0.646; 5-fold PBISC 0.376 / BayesPrism 0.887 / CIBERSORTx 0.592 / scSemiProfiler 0.520; Giroux: 5 donors L1 PBISC 0.256 / BayesPrism 1.455 / CIBERSORTx 0.669; 5-fold PBISC 0.258 / BayesPrism 0.878 / CIBERSORTx 0.591 / scSemiProfiler 0.243", + "source": "data/secS_deconvolution.csv" + } + ], + "original_slide": 46, + "figures_before_palette": [ + "figures/secS_deconv_composition.png" + ] + }, + { + "kind": "content", + "title": "Rare-state frequencies in observed cells", + "section_label": "Fig. 5a", + "layout": "full", + "figures": [ + "figures/review_style_20261013_rare_observed.png" + ], + "keywords": [], + "footnote": "Original five-donor outputs. Median across draws; state share within its parent population. Controls included.", + "footnote_full": "Original five-donor outputs. Median across draws; state share within its parent population. Controls included.", + "sources": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_rare_observed.pdf", + "figures_before_palette": [ + "figures/review_style_20261013_rare_observed.png" + ] + }, + { + "kind": "content", + "title": "Rare-state fidelity: zero-aware reanalysis", + "section_label": "Fig. 5b", + "layout": "full", + "figures": [ + "figures/review_style_20261013_rare_corrected_comparison.png" + ], + "keywords": [], + "footnote": "Original five-donor outputs; zero draws retained and actual donor IDs. 59 state/cohort combinations, including controls.", + "footnote_full": "Original five-donor outputs; zero draws retained and actual donor IDs. 59 state/cohort combinations, including controls.", + "sources": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_rare_corrected_comparison.pdf", + "figures_before_palette": [ + "figures/review_style_20261013_rare_corrected_comparison.png" + ] + }, + { + "kind": "content", + "title": "Behcet: one-gene and multi-gene C1Q rules", + "section_label": "Fig. 5c", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/review_style_20261013_behcet.png" + ], + "keywords": [], + "footnote": "Descriptive: unmatched reference includes training donors. Same three pairs; two identity-pass and one uncertain.", + "footnote_full": "Descriptive: unmatched reference includes training donors. Same three pairs; two identity-pass and one uncertain.", + "sources": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_behcet.pdf", + "figures_before_palette": [ + "figures/review_style_20261013_behcet.png" + ] + }, + { + "kind": "status", + "title": "Remaining work", + "section_label": "Figure status", + "rows": [ + [ + "Protocol updates", + "Total n = 5/10/20/40 without extra validation; exact-five samples in multi-visit cohorts." + ], + [ + "Analysis of existing outputs", + "Patient-aware contrasts, rare-state scoring and healthy cell visualizations." + ], + [ + "Figure 6: new data", + "Reproduce a published finding in the user-prepared bulk-only cohort." + ] + ], + "footnote": "Existing results remain visible. Protocol updates and missing-data work are identified separately." + }, + { + "kind": "content", + "section": "2", + "section_label": "Supporting", + "title": "Control-cell embedding diagnostic", + "layout": "two_figures", + "figures": [ + "figures/sec05_cells_umap.png", + "figures/method_palette_20261013/sec05_cells_metrics.png" + ], + "keywords": [ + "Monocytes: raw 30.6% vs real 16.7%", + "Same noise: 19% of T cells become monocytes", + "Precision: raw 0.36 vs pseudobulk 0.92" + ], + "footnote": "Existing same-noise input diagnostic; not the new LoRA/scSemi OOF embedding. UMAP overlap is not a quantitative score.", + "footnote_full": "Existing same-noise input diagnostic; not the new LoRA/scSemi OOF embedding. UMAP overlap is not a quantitative score.\nOriginal context: ImMicro, 107 donors. Final model f01 6k 1\u00d7 E16 (DDIM 50, EMA); each cell starts from the same noise under every input. Precision/coverage from one scoring call. Composition L1: raw 0.455, pseudobulk 0.166, cohort-mean null 0.161. The UMAP is fitted on real cells and overstates overlap.", + "sources": "tools/sec05_cells_offmanifold.py; data/sec05_cells_metrics.csv, data/sec05_cells_umap_coords.csv", + "refs": [ + "Kynk\u00e4\u00e4nniemi 2019, NeurIPS", + "Naeem 2020, ICML" + ], + "refs_full": [ + "Kynk\u00e4\u00e4nniemi et al. Improved precision and recall metric for assessing generative models. NeurIPS 2019", + "Naeem et al. Reliable fidelity and diversity metrics for generative models. ICML 2020" + ], + "original_slide": 19, + "figures_before_palette": [ + "figures/sec05_cells_umap.png", + "figures/sec05_cells_metrics.png" + ] + }, + { + "kind": "content", + "section": "S", + "section_label": "Supporting", + "sec09_index": 1, + "title": "Generated-cell quality across cohorts", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/sec09_fewpair_cells.png" + ], + "keywords": [ + "LoRA-5 > scSemi-5: coverage, precision 5/5", + "Beats zero-shot + calibration: coverage 5/5", + "Lower L1 than calibration: ImMicro only" + ], + "footnote": "Completed five-donor comparisons. Multi-visit cohorts used more than five paired samples.", + "footnote_full": "Completed five-donor comparisons. Multi-visit cohorts used more than five paired samples.\nOriginal context: Base f01 6k\u00b71x\u00b7E16 (EMA, DDIM 50) + LoRA r=16, 3,000 steps. LoRA-5 / scSemi-5 = LoRA / scSemiProfiler trained on the same 5 donors (5\u201315 pairs); test = all other usable pairs. Zero-shot = base model without fine-tuning, fed each test pair's real bulk as measured; zero-shot + calibration = the same model fed that bulk after gene-wise calibration, alpha fitted on the same 5 training donors' pairs (both in the same call as LoRA-5). scSemi-5 scored in a separate call, same test set and reference (sha-checked); 25/25 draws. Verdict = better in >= 80% of shared draws. LoRA-5 lower L1 than scSemi-5 in 4/5 (Giroux split 3/5). LoRA-5 > zero-shot + calibration: coverage 5/5, precision 3/5, depth-matched 5/5 and 3/5 cohorts; > zero-shot (real bulk): coverage 5/5, precision 5/5, depth-matched 5/5 and 4/5. Depth-matched P/C vs scSemi-5: same ranking, coverage LoRA-5 0.83\u20130.97 vs scSemi-5 0.15\u20130.28. Composition L1, LoRA-5 minus zero-shot + calibration, per-donor Wilcoxon (gain = p < 0.05): ImMicro -0.03 (p 2e-10), Sepsis +0.02 (p 0.79), Imperial +0.02 (p 0.38), Giroux -0.06 (p 0.38), Ji -0.01 (p 0.70); minus zero-shot (real bulk): ImMicro -0.27 (p 2e-18), Sepsis -0.04 (p 0.15), Imperial -0.35 (p 2e-08), Giroux -0.11 (p 0.04), Ji -0.23 (p 0.02). scANVI labels; P/C = PCA-50, k=10 balls. Scores: scores//fewshot_n5/.", + "sources": "tools/sec09_paired_validation.py; data/sec09_fewpair_cells.csv, data/sec09_scsemi_status.csv, data/sec09_slide_text.json", + "numbers": [ + { + "value": "LoRA-5 vs scSemi-5 coverage (shared-draw means; LoRA-better draws/n; verdict): ImMicro 0.792 vs 0.036 (5/5, lora); Sepsis 0.694 vs 0.080 (5/5, lora); Imperial 0.737 vs 0.059 (5/5, lora); Giroux 0.961 vs 0.607 (5/5, lora); Ji 0.707 vs 0.145 (5/5, lora) -> LoRA verdicts 5/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json and draw_scsemi5/summary.json" + }, + { + "value": "LoRA-5 vs scSemi-5 precision (shared-draw means; LoRA-better draws/n; verdict): ImMicro 0.952 vs 0.518 (5/5, lora); Sepsis 0.906 vs 0.728 (5/5, lora); Imperial 0.853 vs 0.330 (5/5, lora); Giroux 1.000 vs 0.992 (5/5, lora); Ji 0.978 vs 0.840 (5/5, lora) -> LoRA verdicts 5/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json and draw_scsemi5/summary.json" + }, + { + "value": "LoRA-5 vs scSemi-5 comp_l1 (shared-draw means; LoRA-better draws/n; verdict): ImMicro 0.176 vs 0.233 (5/5, lora); Sepsis 0.466 vs 0.612 (5/5, lora); Imperial 0.503 vs 0.602 (5/5, lora); Giroux 0.256 vs 0.270 (3/5, split); Ji 0.513 vs 0.592 (4/5, lora) -> LoRA verdicts 4/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json and draw_scsemi5/summary.json" + }, + { + "value": "LoRA-5 vs scSemi-5 coverage_dm (shared-draw means; LoRA-better draws/n; verdict): ImMicro 0.966 vs 0.284 (5/5, lora); Sepsis 0.834 vs 0.158 (5/5, lora); Imperial 0.863 vs 0.152 (5/5, lora); Giroux 0.974 vs 0.238 (5/5, lora); Ji 0.908 vs 0.150 (5/5, lora) -> LoRA verdicts 5/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json and draw_scsemi5/summary.json" + }, + { + "value": "LoRA-5 vs scSemi-5 precision_dm (shared-draw means; LoRA-better draws/n; verdict): ImMicro 0.975 vs 0.829 (5/5, lora); Sepsis 0.914 vs 0.504 (5/5, lora); Imperial 0.881 vs 0.429 (5/5, lora); Giroux 0.995 vs 0.426 (5/5, lora); Ji 0.974 vs 0.564 (5/5, lora) -> LoRA verdicts 5/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json and draw_scsemi5/summary.json" + }, + { + "value": "LoRA-5 vs zero-shot + calibration coverage (same-draw means; LoRA-better draws/n; verdict): ImMicro 0.792 vs 0.535 (5/5, lora); Sepsis 0.694 vs 0.428 (5/5, lora); Imperial 0.737 vs 0.634 (5/5, lora); Giroux 0.961 vs 0.944 (5/5, lora); Ji 0.707 vs 0.114 (5/5, lora) -> LoRA verdicts 5/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json (arms lora5, zs_cal)" + }, + { + "value": "LoRA-5 vs zero-shot + calibration precision (same-draw means; LoRA-better draws/n; verdict): ImMicro 0.952 vs 0.904 (5/5, lora); Sepsis 0.906 vs 0.899 (3/5, split); Imperial 0.853 vs 0.845 (3/5, split); Giroux 1.000 vs 0.996 (4/5, lora); Ji 0.978 vs 0.782 (5/5, lora) -> LoRA verdicts 3/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json (arms lora5, zs_cal)" + }, + { + "value": "LoRA-5 vs zero-shot + calibration comp_l1 (same-draw means; LoRA-better draws/n; verdict): ImMicro 0.176 vs 0.203 (5/5, lora); Sepsis 0.466 vs 0.446 (2/5, split); Imperial 0.503 vs 0.506 (2/5, split); Giroux 0.256 vs 0.288 (2/5, split); Ji 0.513 vs 0.514 (3/5, split) -> LoRA verdicts 1/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json (arms lora5, zs_cal)" + }, + { + "value": "LoRA-5 vs zero-shot + calibration coverage_dm (same-draw means; LoRA-better draws/n; verdict): ImMicro 0.966 vs 0.780 (5/5, lora); Sepsis 0.834 vs 0.658 (5/5, lora); Imperial 0.863 vs 0.681 (5/5, lora); Giroux 0.974 vs 0.937 (5/5, lora); Ji 0.908 vs 0.357 (5/5, lora) -> LoRA verdicts 5/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json (arms lora5, zs_cal)" + }, + { + "value": "LoRA-5 vs zero-shot + calibration precision_dm (same-draw means; LoRA-better draws/n; verdict): ImMicro 0.975 vs 0.940 (5/5, lora); Sepsis 0.914 vs 0.909 (3/5, split); Imperial 0.881 vs 0.868 (3/5, split); Giroux 0.995 vs 0.989 (5/5, lora); Ji 0.974 vs 0.658 (5/5, lora) -> LoRA verdicts 3/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json (arms lora5, zs_cal)" + }, + { + "value": "LoRA-5 vs zero-shot (real bulk) coverage (same-draw means; LoRA-better draws/n; verdict): ImMicro 0.792 vs 0.190 (5/5, lora); Sepsis 0.694 vs 0.076 (5/5, lora); Imperial 0.737 vs 0.114 (5/5, lora); Giroux 0.961 vs 0.472 (5/5, lora); Ji 0.707 vs 0.066 (5/5, lora) -> LoRA verdicts 5/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json (arms lora5, zs_raw)" + }, + { + "value": "LoRA-5 vs zero-shot (real bulk) precision (same-draw means; LoRA-better draws/n; verdict): ImMicro 0.952 vs 0.677 (5/5, lora); Sepsis 0.906 vs 0.625 (5/5, lora); Imperial 0.853 vs 0.500 (5/5, lora); Giroux 1.000 vs 0.980 (5/5, lora); Ji 0.978 vs 0.799 (5/5, lora) -> LoRA verdicts 5/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json (arms lora5, zs_raw)" + }, + { + "value": "LoRA-5 vs zero-shot (real bulk) comp_l1 (same-draw means; LoRA-better draws/n; verdict): ImMicro 0.176 vs 0.445 (5/5, lora); Sepsis 0.466 vs 0.510 (5/5, lora); Imperial 0.503 vs 0.806 (5/5, lora); Giroux 0.256 vs 0.351 (5/5, lora); Ji 0.513 vs 0.721 (5/5, lora) -> LoRA verdicts 5/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json (arms lora5, zs_raw)" + }, + { + "value": "LoRA-5 vs zero-shot (real bulk) coverage_dm (same-draw means; LoRA-better draws/n; verdict): ImMicro 0.966 vs 0.213 (5/5, lora); Sepsis 0.834 vs 0.073 (5/5, lora); Imperial 0.863 vs 0.151 (5/5, lora); Giroux 0.974 vs 0.549 (5/5, lora); Ji 0.908 vs 0.199 (5/5, lora) -> LoRA verdicts 5/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json (arms lora5, zs_raw)" + }, + { + "value": "LoRA-5 vs zero-shot (real bulk) precision_dm (same-draw means; LoRA-better draws/n; verdict): ImMicro 0.975 vs 0.671 (5/5, lora); Sepsis 0.914 vs 0.595 (5/5, lora); Imperial 0.881 vs 0.561 (5/5, lora); Giroux 0.995 vs 0.996 (2/5, split); Ji 0.974 vs 0.480 (5/5, lora) -> LoRA verdicts 4/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json (arms lora5, zs_raw)" + }, + { + "value": "coverage cohort means over all draws: LoRA-5 ImMicro 0.792, Sepsis 0.694, Imperial 0.737, Giroux 0.961, Ji 0.707; scSemi-5 ImMicro 0.036, Sepsis 0.080, Imperial 0.059, Giroux 0.607, Ji 0.145; zero-shot + calibration ImMicro 0.535, Sepsis 0.428, Imperial 0.634, Giroux 0.944, Ji 0.114; zero-shot ImMicro 0.190, Sepsis 0.076, Imperial 0.114, Giroux 0.472, Ji 0.066", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/summary.json and draw_scsemi5/summary.json" + }, + { + "value": "LoRA-5 minus zero-shot + calibration composition L1, per-donor Wilcoxon (draw-averaged): ImMicro -0.027, n=107, p=2.2e-10; Sepsis +0.018, n=82, p=0.79; Imperial +0.016, n=57, p=0.38; Giroux -0.065, n=8, p=0.38; Ji -0.012, n=20, p=0.7 -> gain in 1/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/per_donor.csv; /home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data//folds.csv" + }, + { + "value": "LoRA-5 minus zero-shot (real bulk) composition L1, per-donor Wilcoxon (draw-averaged): ImMicro -0.270, n=107, p=2.2e-18; Sepsis -0.044, n=82, p=0.15; Imperial -0.348, n=57, p=1.6e-08; Giroux -0.107, n=8, p=0.039; Ji -0.232, n=20, p=0.017 -> gain in 4/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores//fewshot_n5/draw/per_donor.csv; /home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data//folds.csv" + }, + { + "value": "scSemiProfiler draws used / LoRA draws: ImMicro 5/5, Sepsis 5/5, Imperial 5/5, Giroux 5/5, Ji 5/5", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00/data/sec09_scsemi_status.csv" + }, + { + "value": "training pairs per draw: ImMicro 5, Sepsis 5, Imperial 12\u201315, Giroux 12\u201314, Ji 7\u20139; test pairs: ImMicro 102, Sepsis 77, Imperial 142\u2013145, Giroux 10\u201312, Ji 24\u201326", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data//fewshot_n5/DRAWS.json (n_train_rows, n_test_rows)" + }, + { + "value": "LoRA r = 16; 3000 fixed steps; base runs/f01_6k_1x_seed20260627/checkpoint_epoch_016.pt; sampling: test donors only, 1,500 cells each, seed 20260927, DDIM 50, EMA, Bernoulli read-out", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/runs/immicro/fewshot_n5/draw0/lora/RUN.json; /home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/README.md" + }, + { + "value": "P/C definition: precision/coverage/density (PCA-50 + k = 10 balls fitted on the scored donors' observed R1 cells); PCA components in every summary.json = 50", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/tools/score_ft.py" + } + ], + "refs": [ + "Kynk\u00e4\u00e4nniemi 2019, NeurIPS", + "Naeem 2020, ICML", + "Wang 2024, Nat Commun" + ], + "refs_full": [ + "Kynk\u00e4\u00e4nniemi et al. Improved precision and recall metric for assessing generative models. NeurIPS 2019", + "Naeem et al. Reliable fidelity and diversity metrics for generative models. ICML 2020", + "Wang et al. scSemiProfiler: advancing large-scale single-cell studies through semi-profiling with deep generative models and active learning. Nat Commun 15:5989 (2024). doi:10.1038/s41467-024-50150-1" + ], + "original_slide": 58, + "figures_before_palette": [ + "figures/sec09_fewpair_cells.png" + ] + }, + { + "kind": "content", + "section": "S", + "section_label": "Supporting", + "text_source": { + "file": "data/sec09_extra_slide_text.json", + "index": 2 + }, + "title": "Precision and coverage across neighbourhood sizes", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/sec09_extra_ksens.png" + ], + "keywords": [ + "Smaller k lowers every set, real included", + "Coverage: LoRA-5 first at all k, 25/25", + "Precision: LoRA-5 first in 20\u201323/25 draws" + ], + "footnote": "Completed sensitivity at k=3,5,10; no new sensitivity campaign required.", + "footnote_full": "Completed sensitivity at k=3,5,10; no new sensitivity campaign required.\nOriginal context: Precision and coverage recomputed with k = 3, 5, 10 nearest-neighbour balls in the scorer's PCA-50 (k = 10 reproduces the main-slide values exactly). Means over the 5 draws per cohort; error bars = SD; grey = real vs real (R1 vs R2, same samples). Both metrics fall with smaller k for every set including the real control (real coverage at k = 3: 0.92\u20130.99), so compare arms with the real row, not with 1. Expected coverage of a perfect generator (Naeem et al. 2020, eq. 9) at k = 3/5/10: 0.88/0.97/1.00 in every cohort. LoRA-5 first on coverage at every k in 25/25 draws; on precision in 23/21/20 of 25 draws (k = 3/5/10); not first: Sepsis 0/1/2, Imperial 2/2/2, Giroux 0/1/1 draws at k = 3/5/10, beaten by zero-shot + calibration. 5-fold pooled design (not drawn): coverage ranking of the 5 arms across k unchanged; LoRA first at every k in 5/5 cohorts. Scores: scores_extra//fewshot_n5/draw/metrics.csv.", + "sources": "tools/sec09_extra_metrics.py; data/sec09_extra_ksens.csv; plans/paired_finetune_intradataset_20260927/scores_extra/", + "numbers": [ + { + "value": "LoRA-5 first on coverage at k = 3, 5 and 10: 25/25 draws", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores_extra//fewshot_n5/draw/{metrics,en}.csv" + }, + { + "value": "LoRA-5 first on precision (draws) at k = 3/5/10: 23/21/20 of 25", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores_extra//fewshot_n5/draw/{metrics,en}.csv" + }, + { + "value": "real vs real coverage at k = 3: ImMicro 0.932, Sepsis 0.925, Imperial 0.919, Giroux 0.993, Ji 0.980", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores_extra//fewshot_n5/draw/{metrics,en}.csv" + }, + { + "value": "every set falls with smaller k (cohort means): True", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores_extra//fewshot_n5/draw/{metrics,en}.csv" + } + ], + "refs": [ + "Naeem 2020, ICML", + "Khayatkhoei 2023, ICML" + ], + "refs_full": [ + "Naeem et al. Reliable fidelity and diversity metrics for generative models. ICML 2020", + "Khayatkhoei & AbdAlmageed. Emergent asymmetry of precision and recall for measuring fidelity and diversity of generative models in high dimensions. ICML 2023" + ], + "original_slide": 60, + "figures_before_palette": [ + "figures/sec09_extra_ksens.png" + ] + }, + { + "kind": "content", + "section": "S", + "section_label": "Supporting", + "text_source": { + "file": "data/sec09_same_pairs_slide_text.json", + "index": 0 + }, + "title": "Same training pairs: five-fold comparison", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/sec09_same_pairs.png" + ], + "keywords": [ + "Coverage, precision: LoRA ahead in 5/5", + "LoRA ahead in composition 4/5, logFC r 3/4", + "19\u201362\u00d7 less GPU time per fold" + ], + "footnote": "Completed original five-fold LoRA/scSemi comparison; not the new total-n curve.", + "sources": "tools/sec09_same_pairs.py; data/sec09_same_pairs.csv; plans/paired_finetune_intradataset_20260927/scsemi/analysis/cv_budget_20261005/", + "say": "Here scSemiProfiler gets exactly the pairs LoRA trains on, every training donor of the fold. LoRA is still ahead on every metric and about 20 times cheaper. Sepsis, Imperial and Giroux are still running.", + "footnote_full": "Completed original five-fold LoRA/scSemi comparison; not the new total-n curve.\nOriginal context: 5-fold CV: for every fold, scSemiProfiler's representatives are exactly LoRA's training + inner-val pairs (the other four folds; checked per fold), test = the fold's samples from their real bulk only. Pooled = one scoring call per method over every usable sample, same donors and reference (sha-checked). scSemiProfiler = upstream c2aea02, default settings, official output. GPU time per fold on one GB10: LoRA = fine-tune + sampling 1,500 cells per test sample; scSemiProfiler = pretraining of every representative + one model per test sample (receipts); data preparation and scoring not counted on either side. scSemiProfiler ahead on composition: Giroux. scSemiProfiler ahead on logFC r: Giroux. Equal pair budget, ImMicro (n = 5, 10, 20, 40 pairs): scSemiProfiler L1 0.218\u20130.227 and coverage 0.04\u20130.04 at every n; LoRA L1 0.171\u20130.188, coverage 0.76\u20130.84 (drawn on the 'Pairs needed: composition, coverage, training steps' slide). Equal pair budget, Sepsis (n = 5, 10, 20, 40 pairs): scSemiProfiler L1 0.516\u20130.536 and coverage 0.10\u20130.10 at every n; LoRA L1 0.365\u20130.428, coverage 0.77\u20130.82 (drawn on the 'Pairs needed: composition, coverage, training steps' slide). Source: plans/paired_finetune_intradataset_20260927/scores//{pooled,scsemi_cv_20261002/pooled}/ and scsemi/analysis/cv_budget_20261005/.", + "numbers": [ + { + "value": "ImMicro: L1 LoRA 0.162 / scSemi 0.217; coverage 0.784 / 0.038; precision 0.924 / 0.505; logFC r nan / nan; top-50 DEG None / None; GPU per fold LoRA 14.7 min / scSemi 4.75 h", + "source": "data/sec09_same_pairs.csv" + }, + { + "value": "Sepsis: L1 LoRA 0.362 / scSemi 0.502; coverage 0.829 / 0.096; precision 0.926 / 0.760; logFC r 0.527 / 0.435; top-50 DEG 23 / 6; GPU per fold LoRA 7.4 min / scSemi 7.69 h", + "source": "data/sec09_same_pairs.csv" + }, + { + "value": "Imperial: L1 LoRA 0.376 / scSemi 0.520; coverage 0.850 / 0.082; precision 0.847 / 0.348; logFC r 0.819 / 0.705; top-50 DEG 143 / 87; GPU per fold LoRA 6.2 min / scSemi 6.18 h", + "source": "data/sec09_same_pairs.csv" + }, + { + "value": "Giroux: L1 LoRA 0.258 / scSemi 0.243; coverage 0.928 / 0.354; precision 0.998 / 0.872; logFC r 0.461 / 0.537; top-50 DEG 5 / 0; GPU per fold LoRA 3.6 min / scSemi 2.89 h", + "source": "data/sec09_same_pairs.csv" + }, + { + "value": "Ji: L1 LoRA 0.504 / scSemi 0.575; coverage 0.857 / 0.167; precision 0.971 / 0.858; logFC r 0.479 / 0.273; top-50 DEG 15 / 0; GPU per fold LoRA 4.4 min / scSemi 1.48 h", + "source": "data/sec09_same_pairs.csv" + }, + { + "value": "Equal pair budget, ImMicro (n = 5, 10, 20, 40 pairs): scSemiProfiler L1 0.218\u20130.227 and coverage 0.04\u20130.04 at every n; LoRA L1 0.171\u20130.188, coverage 0.76\u20130.84 (drawn on the 'Pairs needed: composition, coverage, training steps' slide). Equal pair budget, Sepsis (n = 5, 10, 20, 40 pairs): scSemiProfiler L1 0.516\u20130.536 and coverage 0.10\u20130.10 at every n; LoRA L1 0.365\u20130.428, coverage 0.77\u20130.82 (drawn on the 'Pairs needed: composition, coverage, training steps' slide).", + "source": "data/sec09_same_pairs_budget.csv" + } + ], + "original_slide": 62, + "figures_before_palette": [ + "figures/sec09_same_pairs.png" + ] + }, + { + "kind": "content", + "section": "S", + "section_label": "Supporting", + "layout": "full", + "sources": "tools/sec08_cross_cohort.py; data/sec08_cross_cohort.csv, data/sec08_cross_cohort_steps.csv; plans/paired_crosscohort_20261002/ (README, scores/aggregate.csv)", + "text_source": { + "file": "data/sec08_cross_cohort_slide_text.json", + "index": 0 + }, + "title": "Cross-cohort pairs do not replace local pairs", + "figures": [ + "figures/method_palette_20261013/sec08_cross_cohort.png" + ], + "keywords": [ + "Train on the other cohorts, test the held-out one", + "Calibration ties or beats them on composition in 4/5", + "Own-cohort pairs best on coverage 5/5, disease r 4/4" + ], + "footnote": "Completed leave-one-cohort-out test at K=1000; target-cohort sc was excluded from training.", + "say": "Why one LoRA per cohort: a model trained on the other PBMC cohorts' pairs (three, or four for Ji, plus Liu 2021) rarely beats a zero-pair gene-wise calibration on composition, and no arm without the cohort's own pairs matches the own-cohort LoRA on coverage or disease signal.", + "footnote_full": "Completed leave-one-cohort-out test at K=1000; target-cohort sc was excluded from training.\nOriginal context: Each held-out cohort is scored on all its usable samples against the same frozen reference as its within-cohort benchmark. Zero-shot = base model on real bulk. Calibration from other cohorts = gene-wise alpha (mean pseudobulk CPM / mean bulk CPM) fitted on the training cohorts' pairs, cohort-balanced; 0 target pairs. LoRA r16 / r64 / full fine-tune = trained on the other PBMC cohorts' pairs plus Liu 2021 (training only: its cells are in the pretraining corpus; 44 pairs after excluding 14 hashtag-negative-heavy rows), cohort-balanced batches, inner-validation on 10 % of each training cohort's donors, step-saturation protocol (20k-step blocks, stop when a block lowers the inner-validation loss by < 1 % of the total), value at the last milestone (number in the bar = steps). Ji's bulk is whole blood and is never trained on. Own-cohort LoRA = LoRA r16 on the cohort's own pairs, 5-fold pooled. Giroux's logFC r excludes 196 sex-linked genes. Best on composition among zero-target-pair arms: ImMicro Calibration from other cohorts, Sepsis Calibration from other cohorts, Imperial Full fine-tune, other cohorts, Giroux Calibration from other cohorts, Ji (whole blood) Calibration from other cohorts.", + "numbers": [ + { + "value": "ImMicro: L1 zs 0.446 / xcal 0.172 / lora16sat 0.175 / lora64sat 0.199 / fullsat 0.213 / own 0.162; coverage zs 0.19 / xcal 0.57 / lora16sat 0.45 / lora64sat 0.38 / fullsat 0.35 / own 0.78", + "source": "data/sec08_cross_cohort.csv" + }, + { + "value": "Sepsis: L1 zs 0.509 / xcal 0.444 / lora16sat 0.491 / lora64sat 0.570 / fullsat 0.585 / own 0.362; coverage zs 0.07 / xcal 0.46 / lora16sat 0.34 / lora64sat 0.29 / fullsat 0.35 / own 0.83; logFC r zs -0.02 / xcal 0.10 / lora16sat 0.10 / lora64sat 0.23 / fullsat 0.07 / own 0.53", + "source": "data/sec08_cross_cohort.csv" + }, + { + "value": "Imperial: L1 zs 0.803 / xcal 0.651 / lora16sat 0.635 / lora64sat 0.594 / fullsat 0.580 / own 0.376; coverage zs 0.11 / xcal 0.64 / lora16sat 0.68 / lora64sat 0.72 / fullsat 0.61 / own 0.85; logFC r zs 0.64 / xcal 0.69 / lora16sat 0.49 / lora64sat 0.58 / fullsat 0.46 / own 0.82", + "source": "data/sec08_cross_cohort.csv" + }, + { + "value": "Giroux: L1 zs 0.337 / xcal 0.202 / lora16sat 0.431 / lora64sat 0.410 / fullsat 0.465 / own 0.258; coverage zs 0.29 / xcal 0.89 / lora16sat 0.82 / lora64sat 0.81 / fullsat 0.78 / own 0.93; logFC r zs 0.34 / xcal 0.24 / lora16sat 0.30 / lora64sat 0.28 / fullsat 0.20 / own 0.46", + "source": "data/sec08_cross_cohort.csv" + }, + { + "value": "Ji (whole blood): L1 zs 0.725 / xcal 0.565 / lora16sat 0.740 / lora64sat 0.620 / fullsat 1.030 / own 0.504; coverage zs 0.06 / xcal 0.07 / lora16sat 0.34 / lora64sat 0.40 / fullsat 0.08 / own 0.86; logFC r zs 0.07 / xcal 0.13 / lora16sat 0.05 / lora64sat 0.16 / fullsat 0.14 / own 0.48", + "source": "data/sec08_cross_cohort.csv" + } + ], + "original_slide": 68, + "figures_before_palette": [ + "figures/sec08_cross_cohort.png" + ] + }, + { + "kind": "content", + "section": "S", + "section_label": "Supporting", + "layout": "full", + "sources": "tools/sec08_cross_cohort.py; data/sec08_cross_cohort.csv (minutes = RUN.json wall_seconds / 60), data/sec08_cross_cohort_steps.csv; plans/paired_crosscohort_20261002/ (README, scores/aggregate.csv, runs/hoL_/K1000//RUN.json)", + "text_source": { + "file": "data/sec08_cross_cohort_slide_text.json", + "index": 1 + }, + "title": "Longer cross-cohort training leaves a gap", + "figures": [ + "figures/method_palette_20261013/sec08_cross_cohort_steps.png" + ], + "keywords": [ + "To saturation: r16 20\u201331 min, full 59\u2013107 min", + "Full fine-tune drifts in Sepsis and Ji (Ji L1 0.84 \u2192 1.03)", + "So PBISC fits LoRA on a few pairs per cohort" + ], + "footnote": "Existing saturation experiment; no additional saturation run introduced here.", + "say": "Each run trained until its inner-validation loss saturated. More steps on other cohorts never closed the coverage gap to the cohort's own pairs: Imperial and Giroux improved a little, and full fine-tuning drifted further in Sepsis and Ji (whole-blood bulk).", + "footnote_full": "Existing saturation experiment; no additional saturation run introduced here.\nOriginal context: Same runs as the previous slide. Each point = the milestone checkpoint at the end of a 20k-step block (or at the early stop), sampled and scored on the held-out cohort. Dashed = calibration from other cohorts; dotted = LoRA on the cohort's own pairs. Training stopped (steps): LoRA r16 ImMicro 28k, Sepsis 26k, Imperial 40k, Giroux 35k, Ji 40k; LoRA r64 ImMicro 39k, Sepsis 59k, Imperial 69k, Giroux 69k, Ji 51k; Full fine-tune ImMicro 65k, Sepsis 108k, Imperial 93k, Giroux 120k, Ji 86k. Saturation steps across cohorts: r16 up to 40k, r64 up to 69k, full up to 120k. Training time to saturation (RUN.json wall seconds, one GB10): LoRA r16 ImMicro 21 min, Sepsis 20 min, Imperial 30 min, Giroux 28 min, Ji 31 min; LoRA r64 ImMicro 27 min, Sepsis 44 min, Imperial 54 min, Giroux 53 min, Ji 41 min; Full fine-tune ImMicro 59 min, Sepsis 93 min, Imperial 82 min, Giroux 107 min, Ji 77 min. Gap = no milestone of any mode reaches the own-cohort LoRA's coverage (true in all five cohorts); full fine-tune composition L1 from the first to the last milestone: ImMicro +0.005, Sepsis +0.087, Imperial -0.052, Giroux -0.078, Ji +0.192 (drift >= 0.05 in Sepsis and Ji; improved by >= 0.05 in Imperial and Giroux).", + "numbers": [ + { + "value": "ImMicro LoRA r16: 20k L1 0.163 cov 0.45, 28k L1 0.175 cov 0.45", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "ImMicro LoRA r64: 20k L1 0.196 cov 0.44, 39k L1 0.199 cov 0.38", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "ImMicro Full fine-tune: 20k L1 0.208 cov 0.39, 40k L1 0.211 cov 0.37, 60k L1 0.223 cov 0.35, 65k L1 0.213 cov 0.35", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Sepsis LoRA r16: 20k L1 0.500 cov 0.35, 26k L1 0.491 cov 0.34", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Sepsis LoRA r64: 20k L1 0.550 cov 0.29, 40k L1 0.564 cov 0.29, 59k L1 0.570 cov 0.29", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Sepsis Full fine-tune: 20k L1 0.498 cov 0.33, 40k L1 0.530 cov 0.33, 60k L1 0.546 cov 0.34, 80k L1 0.576 cov 0.34, 100k L1 0.577 cov 0.35, 108k L1 0.585 cov 0.35", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Imperial LoRA r16: 20k L1 0.657 cov 0.68, 40k L1 0.635 cov 0.68", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Imperial LoRA r64: 20k L1 0.622 cov 0.68, 40k L1 0.603 cov 0.71, 60k L1 0.577 cov 0.72, 69k L1 0.594 cov 0.72", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Imperial Full fine-tune: 20k L1 0.632 cov 0.60, 40k L1 0.609 cov 0.60, 60k L1 0.565 cov 0.61, 80k L1 0.576 cov 0.60, 93k L1 0.580 cov 0.61", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Giroux LoRA r16: 20k L1 0.416 cov 0.79, 35k L1 0.431 cov 0.82", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Giroux LoRA r64: 20k L1 0.491 cov 0.80, 40k L1 0.469 cov 0.82, 60k L1 0.430 cov 0.82, 69k L1 0.410 cov 0.81", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Giroux Full fine-tune: 20k L1 0.543 cov 0.75, 40k L1 0.532 cov 0.75, 60k L1 0.499 cov 0.76, 80k L1 0.466 cov 0.76, 100k L1 0.490 cov 0.76, 120k L1 0.465 cov 0.78", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Ji (whole blood) LoRA r16: 20k L1 0.757 cov 0.27, 40k L1 0.740 cov 0.34", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Ji (whole blood) LoRA r64: 20k L1 0.639 cov 0.38, 40k L1 0.668 cov 0.39, 51k L1 0.620 cov 0.40", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Ji (whole blood) Full fine-tune: 20k L1 0.838 cov 0.17, 40k L1 1.006 cov 0.11, 60k L1 1.032 cov 0.08, 80k L1 1.052 cov 0.07, 86k L1 1.030 cov 0.08", + "source": "data/sec08_cross_cohort_steps.csv" + }, + { + "value": "Training time to saturation, LoRA r16: ImMicro 21 min, Sepsis 20 min, Imperial 30 min, Giroux 28 min, Ji 31 min", + "source": "data/sec08_cross_cohort.csv (minutes = RUN.json wall_seconds / 60)" + }, + { + "value": "Training time to saturation, LoRA r64: ImMicro 27 min, Sepsis 44 min, Imperial 54 min, Giroux 53 min, Ji 41 min", + "source": "data/sec08_cross_cohort.csv (minutes = RUN.json wall_seconds / 60)" + }, + { + "value": "Training time to saturation, Full fine-tune: ImMicro 59 min, Sepsis 93 min, Imperial 82 min, Giroux 107 min, Ji 77 min", + "source": "data/sec08_cross_cohort.csv (minutes = RUN.json wall_seconds / 60)" + } + ], + "original_slide": 69, + "figures_before_palette": [ + "figures/sec08_cross_cohort_steps.png" + ] + }, + { + "kind": "content", + "section": "S", + "section_label": "Supporting", + "text_source": { + "file": "data/sec07_no_pretraining_slide_text.json", + "index": 0 + }, + "title": "From-scratch control: evidence and limits", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/sec07_no_pretraining.png" + ], + "keywords": [ + "Same realism; 5 pairs: donor r 0.70 vs 0.17", + "58 pairs: scratch logFC r 0.58 vs 0.49", + "Scratch 61\u201386 min, LoRA 3\u20135 min" + ], + "footnote": "Completed Sepsis fold0 pilot, not all folds. Pretraining supports speed/few-pair adaptation, not universal superiority.", + "sources": "tools/sec07_no_pretraining.py; data/sec07_no_pretraining.csv; plans/paired_finetune_intradataset_20260927/scores/ye2026_sepsis/fold0_scratch_20261005/", + "say": "Pilot on one sepsis fold. Training the same model from scratch on the cohort's pairs gives realistic cells, and with 58 pairs a higher disease logFC r, but with few pairs it cannot place donors: what pretraining buys is the donor-specific composition and minutes instead of an hour per cohort. Folds 1-4 are needed before any claim.", + "footnote_full": "Completed Sepsis fold0 pilot, not all folds. Pretraining supports speed/few-pair adaptation, not universal superiority.\nOriginal context: Pilot on one fold: sepsis fold 0, 17 test donors (12 sepsis, 5 controls); folds 1-4 not run, so no claim yet. From scratch = the same denoiser (54.2 M weights) re-initialised at random (tools/ft_sdd_aug.py --init random; seed 20260627) and trained in full on the cohort's own pairs at the pretraining lr 2e-4, batch 256, EMA 0.999, inner-val selection every 1,000 steps, patience 10, cap 100k steps; the checkpoint supplies only the architecture, 6k panel, value scaler and bulk z-score constants. Pretrained + LoRA = f01 6k 1x E16 + LoRA r16 (lr 5e-4). Same training pairs (5 or 58, + 7-8 inner-val), real-bulk conditions K = 1,000, test donors and sampling seed; recipes = K-diversified pseudobulk augmentation. The 58-pair scratch arms were still improving at the 100k cap. Coverage = PCA-50 k = 10 balls (precision, C2ST, MMD, E_N agree; scores_extra/ye2026_sepsis/fold0_scratch_20261005/). Donor composition r = per cell type (T, B, Mono, NK) Pearson r over the 17 donors of generated vs observed fraction, mean of the four. Dashed = zero-shot. Training time = RUN.json wall seconds on one GB10 (LoRA also needs the one-time 15.9 h pretraining, which every cohort reuses). Source: plans/paired_finetune_intradataset_20260927/scores/ye2026_sepsis/fold0_scratch_20261005/ and README 'No-pretraining ablation'.", + "numbers": [ + { + "value": "5 pairs: pretrained+LoRA L1 0.363 / scratch 0.448; coverage 0.81 / 0.80; logFC r 0.23 / 0.18; donor composition r 0.70 / 0.17; time 3.2 / 61 min; steps (selected/run) 2750/4750 vs 66000/76000", + "source": "data/sec07_no_pretraining.csv" + }, + { + "value": "58 pairs: pretrained+LoRA L1 0.373 / scratch 0.398; coverage 0.86 / 0.86; logFC r 0.49 / 0.58; donor composition r 0.77 / 0.68; time 5.4 / 79 min; steps (selected/run) 6500/8500 vs 100000/100000", + "source": "data/sec07_no_pretraining.csv" + }, + { + "value": "58 + recipes pairs: pretrained+LoRA L1 0.385 / scratch 0.385; coverage 0.87 / 0.88; logFC r 0.45 / 0.54; donor composition r 0.80 / 0.77; time 4.5 / 86 min; steps (selected/run) 3750/5750 vs 100000/100000", + "source": "data/sec07_no_pretraining.csv" + } + ], + "original_slide": 70, + "figures_before_palette": [ + "figures/sec07_no_pretraining.png" + ] + }, + { + "kind": "content", + "section": "A", + "section_label": "Supporting", + "text_source": { + "file": "data/secA_scsemi_readout_slide_text.json", + "index": 0 + }, + "title": "scSemiProfiler readout sensitivity", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/secA_scsemi_readout.png" + ], + "keywords": [ + "Official output zeroes values \u2264 10 (fixed)", + "Removed: scSemi coverage up in 5/5 cohorts", + "No LoRA-5 comparison flips (23/23 same)" + ], + "footnote": "Completed modified-readout check; main results use official output. Models were not retrained for the sensitivity.", + "say": "Appendix only (sensitivity check). The threshold is part of scSemiProfiler's official output, so removing it is a modified run; the main comparison stays the official output. This slide shows that our conclusions do not depend on that read-out.", + "sources": "plans/paired_finetune_intradataset_20260927/scsemi/tools/{resample_output_variants,compare_output_variants}.py; tools/secA_scsemi_readout.py; data/secA_scsemi_readout.csv", + "footnote_full": "Completed modified-readout check; main results use official output. Models were not retrained for the sensitivity.\nOriginal context: Appendix, sensitivity only. scSemiProfiler's official output (upstream commit c2aea02) is the decoder's ZINB mean with every value <= 10 normalized units set to 0; the cut is fixed in the code (vaegan.myVAE.sample, bound = 10, and fast_semi: xsemi * (xsemi > 10)), not an option. At our cohorts' depth (about 1,000-2,000 UMI per cell on the panel) it removes most low- and mid-expressed genes. Here the same trained per-target models were read out again without that cut (CPU, no retraining, target seeds as in the run), exported and scored exactly as the official arm (same 25 five-donor draws, donors, reference; Giroux disease metrics without sex-linked genes). A control re-read with the cut reproduces the official output: pseudobulk r >= 0.996 over 1790 targets, coverage within 0.015. Because this changes the method's official read-out, the main slides keep the official output. Bars = mean of 5 draws, dots = draws. Source: plans/paired_finetune_intradataset_20260927/scsemi/analysis/output_variants_20261003/ (README section 'Read-out sensitivity').", + "numbers": [ + { + "value": "ImMicro: coverage LoRA-5 0.79, official 0.04, no threshold 0.23; precision 0.95 / 0.52 / 0.74; L1 0.176 / 0.233 / 0.199", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scsemi/analysis/output_variants_20261003/per_draw.csv" + }, + { + "value": "Sepsis: coverage LoRA-5 0.69, official 0.08, no threshold 0.18; precision 0.91 / 0.73 / 0.77; L1 0.466 / 0.612 / 0.591; logFC r 0.35 / 0.40 / 0.46; top-50 DEG 16.4 / 5.2 / 14.0", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scsemi/analysis/output_variants_20261003/per_draw.csv" + }, + { + "value": "Imperial: coverage LoRA-5 0.74, official 0.06, no threshold 0.42; precision 0.85 / 0.33 / 0.72; L1 0.503 / 0.602 / 0.603; logFC r 0.75 / 0.56 / 0.66; top-50 DEG 113.6 / 46.8 / 94.6", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scsemi/analysis/output_variants_20261003/per_draw.csv" + }, + { + "value": "Giroux: coverage LoRA-5 0.96, official 0.61, no threshold 0.83; precision 1.00 / 0.99 / 1.00; L1 0.256 / 0.270 / 0.273; logFC r 0.30 / 0.40 / 0.45; top-50 DEG 0.2 / 0.0 / 0.0", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scsemi/analysis/output_variants_20261003/per_draw.csv" + }, + { + "value": "Ji: coverage LoRA-5 0.71, official 0.15, no threshold 0.53; precision 0.98 / 0.84 / 0.92; L1 0.513 / 0.592 / 0.575; logFC r 0.33 / 0.15 / 0.19; top-50 DEG 0.2 / 0.0 / 0.0", + "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scsemi/analysis/output_variants_20261003/per_draw.csv" + } + ], + "original_slide": 74, + "figures_before_palette": [ + "figures/secA_scsemi_readout.png" + ] + }, + { + "kind": "content", + "section": "4", + "section_label": "Supporting", + "title": "SLE pseudobulk development benchmark", + "layout": "full", + "figures": [ + "figures/method_palette_20261013/sec04_sle_benchmark.png" + ], + "keywords": [ + "42 held-out donors, sc-pseudobulk input", + "Beats scSemiProfiler on all four metrics", + "Beyond bulk: partial r 0.21 vs 0.17" + ], + "footnote": "Completed pseudobulk benchmark; separate from external paired real-bulk validation. PBISC five seeds, scSemi one run.", + "sources": "tools/sec04_sle_benchmark.py; data/sec04_sle_benchmark.csv, data/sec04_sle_numbers.csv", + "footnote_full": "Completed pseudobulk benchmark; separate from external paired real-bulk validation. PBISC five seeds, scSemi one run.\nOriginal context: GSE135779, 42 held-out donors (30 SLE / 12 HC). PBISC f01 6k E16, 5 seeds (mean \u00b1 SD); baselines 1 draw. scSemiProfiler also gets real sc of 14 in-cohort donors (1,500 cells each). Partial r: both logFCs residualised on the input bulk's logFC (the cell-type signal the bulk does not carry), 1,006 genes: PBISC 0.208 \u00b1 0.012, scSemiProfiler 0.167.", + "refs": [ + "Wang 2024, Nat Commun", + "Xu 2021, Mol Syst Biol" + ], + "refs_full": [ + "Wang et al. scSemiProfiler: advancing large-scale single-cell studies through semi-profiling with deep generative models and active learning. Nat Commun 15:5989 (2024). doi:10.1038/s41467-024-50150-1", + "Xu et al. Probabilistic harmonization and annotation of single-cell transcriptomics data with deep generative models. Mol Syst Biol 17:e9620 (2021)" + ], + "original_slide": 42, + "figures_before_palette": [ + "figures/sec04_sle_benchmark.png" + ] + } + ] +} diff --git a/paper-releases/pbisc-diffusion/updates/20261013T022035Z-pbisc-red-scsemi-blue/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_healthy_composition.pdf b/paper-releases/pbisc-diffusion/updates/20261013T022035Z-pbisc-red-scsemi-blue/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_healthy_composition.pdf new file mode 100644 index 0000000000000000000000000000000000000000..799ae1038129fd02abb29ac7761a6964d9a8c59e Binary files /dev/null and b/paper-releases/pbisc-diffusion/updates/20261013T022035Z-pbisc-red-scsemi-blue/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_healthy_composition.pdf differ diff --git a/paper-releases/pbisc-diffusion/updates/20261013T022035Z-pbisc-red-scsemi-blue/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_healthy_composition.svg b/paper-releases/pbisc-diffusion/updates/20261013T022035Z-pbisc-red-scsemi-blue/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_healthy_composition.svg new file mode 100644 index 0000000000000000000000000000000000000000..52387a4df65e1d9b0641a982d6886ce30b99836c --- /dev/null +++ b/paper-releases/pbisc-diffusion/updates/20261013T022035Z-pbisc-red-scsemi-blue/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/figures/review_style_20261013_healthy_composition.svg @@ -0,0 +1,645 @@ + + + + + + + + 2026-10-13T11:17:27.614530 + image/svg+xml + + + Matplotlib v3.10.7, https://matplotlib.org/ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + 0 + + + + + + + + + + + + + 2 + + + + + + + + + + + + + 4 + + + + + + + + + + + + + 6 + + + + + + + + + + + + + 8 + + + + Mean absolute error (percentage points) + + + + + + + T cells + + + + + + B cells + + + + + + Monocytes + + + + + + NK cells + + + + + + DC + + + + + + Other + + + + + + + + + + + + + + + + + + + + + + + + + + 6.22 + + + 1.95 + + + 4.55 + + + 3.64 + + + 0.81 + + + 0.40 + + + + + + + + + + + + + + + + + + + + + 6.86 + + + 3.17 + + + 5.39 + + + 4.42 + + + 3.02 + + + 0.38 + + + Error by cell type ↓ + + + + + + + + + + + + + + + + + + 0 + + + + + + + + + + 25 + + + + + + + + + + 50 + + + + + + + + + + 75 + + + + + + + + + + 100 + + + + Donors with lower error (%) + + + + + + + T cells + + + + + + B cells + + + + + + Monocytes + + + + + + NK cells + + + + + + DC + + + + + + Other + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + 63 + + + 44 + + + 93 + + + 14 + + + 61 + + + 46 + + + 78 + + + 29 + + + 101 + + + 6 + + + 36 + + + 71 + + + Which method is closer? + + + + Overall composition L1: 0.176 vs 0.232 + + + PBISC: 24.4% lower mean error; closer in 93/107 donors + + + Absolute error per draw, then averaged within donor. Equal donor weights; descriptive comparison. + + + Numbers in bars = donors (n = 107) + + + + + + + PBISC LoRA + + + + + + scSemiProfiler + + + + + + + + + + + +