Archive 4 checkpoints (1-4)

.gitattributes

@@ -102,3 +102,5 @@ paper-releases/pbisc-diffusion/updates/20261008T022410Z-deconv-scsemi-deck/diffu
 paper-releases/pbisc-diffusion/updates/20261008T022410Z-deconv-scsemi-deck/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/PBISC-Diffusion_paper_summary_20260930.pptx filter=lfs diff=lfs merge=lfs -text
 paper-releases/pbisc-diffusion/updates/20261008T050202Z-scsemi-imperial-deck/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/PBISC-Diffusion_paper_summary_20260930.pdf filter=lfs diff=lfs merge=lfs -text
 paper-releases/pbisc-diffusion/updates/20261008T050202Z-scsemi-imperial-deck/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/PBISC-Diffusion_paper_summary_20260930.pptx filter=lfs diff=lfs merge=lfs -text
+paper-releases/pbisc-diffusion/updates/20261008T060837Z-deck-new-figure-split-time/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/PBISC-Diffusion_paper_summary_20260930.pdf filter=lfs diff=lfs merge=lfs -text
+paper-releases/pbisc-diffusion/updates/20261008T060837Z-deck-new-figure-split-time/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/PBISC-Diffusion_paper_summary_20260930.pptx filter=lfs diff=lfs merge=lfs -text

paper-releases/pbisc-diffusion/updates/20261008T060837Z-deck-new-figure-split-time/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/PBISC-Diffusion_paper_summary_20260930.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:486e79f898c1787f15b587a8c45917824e799191e86e17cb42c093419388835a
+size 9495220

paper-releases/pbisc-diffusion/updates/20261008T060837Z-deck-new-figure-split-time/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/PBISC-Diffusion_paper_summary_20260930.pptx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e5621810db8844be7a617b204d5bc2788a48d35e93e83067d487643c142b5c75
+size 10427532

paper-releases/pbisc-diffusion/updates/20261008T060837Z-deck-new-figure-split-time/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/README.md

@@ -0,0 +1,260 @@
+# PBISC-Diffusion paper summary deck (2026-09-30, r00; ordered by the six manuscript figures since 2026-10-06)
+
+65-slide English deck, ordered by the six main figures of the Nature-style manuscript plan (user request 2026-10-06,
+following `paper-release-20260928/reference-papers/Nature Style Manuscript Writing.pdf`):
+- Title and outline: 6 figures, Supplementary, Appendix, Discussion.
+- "Manuscript plan: six main figures": a table of each figure's claim and how many panels are ready.
+- For each figure:
+  - a divider;
+  - a plan slide, a table of panel / content / evidence slide / status (Ready, Adapt, Draft, Build, Run, Plan);
+  - the existing slides that serve its panels, each labelled "Fig. 2c" etc. in the top-right corner.
+- Supplementary: a plan slide plus the from-scratch results. Appendix: the modified scSemiProfiler run. Discussion: the
+  other-tissues outlook.
+
+User decisions of 2026-10-06:
+- No PBISC-VAE: scSemiProfiler, a VAE-based method, is the comparison (Fig. 3).
+- Fig. 2 shows LoRA first on the healthy cohort (ImMicro) with the realism metrics (precision, coverage, classifier
+  AUROC, MMD, E_N); the disease-cohort comparison comes later.
+- CYBERSORTx and other bulk deconvolution methods go to the Supplementary only (PBISC generates single cells).
+- Figs 4 and 6 stay unplanned. Their existing slides are kept in the section as "Fig. 4 candidate" / "Fig. 6 candidate".
+
+Only order, section labels, the agenda and new plan or draft slides changed. The new slides are "Overview draft (Nature
+style)", the manuscript's candidate Fig. 1, and "Pretraining corpus", the Fig. 1d strip. No new result figures were
+made.
+- The plan lives in `data/figure_plan.json`.
+- `tools/figure_plan.py` draws the plan tables and the slide text, and checks that every slide is placed exactly once.
+- `tools/restructure_figure_order_20261006.py` rebuilds the order from the plan. It is rerun after every plan change and
+  keeps all slide fields.
+- The earlier section order (10 sections, 45 slides) is in `_archive/2026-10-06_before_figure_order/`.
+
+Short keyword lines only; the figures carry the results. Every slide that uses a published method, metric or
+baseline has a one-line "References:" footer; full citations are in the speaker notes.
+
+- `PBISC-Diffusion_paper_summary_20260930.pptx`: the deck
+- `PBISC-Diffusion_paper_summary_20260930.pdf`: LibreOffice PDF of the deck
+- `data/slides_spec.json`: slide order, titles, keywords, footnotes, references, speaker-note text
+- `generate_deck.py`: builds the pptx from `data/slides_spec.json` only (python-pptx)
+- `figures/`: exactly the figures the slides show (PNG + PDF), made by `tools/sec*.py`. Intermediate panels go to a
+  temporary folder; superseded figures are deleted, not kept (2026-10-02, user request; every deleted file was
+  checked to be in the paper archive by sha256 first).
+- `data/sec*.csv` and `data/*_slide_text.json`: every plotted or quoted number with its source path
+- `render/`: PDF and 70-dpi slide JPGs for checking
+
+## Slide order
+
+| section | slides |
+|---|---|
+| 1 Model overview | bulk in, cells out; denoiser |
+| 2 Backbone | SDD → PBISC-Diffusion; bulk conditioning |
+| 3 Architecture ablation | width & context (loss, benchmark); gene panel |
+| 4 Metrics + SLE benchmark | **how we score generated cells** (metric table with references); SLE benchmark; per cell type; bulk broadcast baseline |
+| 5 Real bulk vs pseudobulk | input shift; same shift in three cohorts; cells off-manifold; disease direction flips |
+| 6 Paired fine-tuning | method; evaluation design |
+| 7 LoRA ablations | full vs LoRA vs cond.-encoder-only; LoRA rank r = 4 / 16 / 64; K diversification |
+| 8 Pairs needed | composition / coverage / steps (with scSemiProfiler at the same n); donor and disease signal; pairs from other cohorts; longer training on them (with training minutes) |
+| 9 Paired validation | cohorts; 5 donors: cells, realism beyond P/C, P/C at k = 3 / 5 / 10, scSemiProfiler copies its representative, disease signal, CAT curves, time to apply; 5-fold CV (with scSemiProfiler on the same pairs); same training pairs: LoRA vs scSemiProfiler; 5-fold realism metrics |
+| 10 Next steps | paired-subset cohorts (planned); other tissues from a few dozen pairs (Discussion framing of the from-scratch results) |
+| Supplementary | no pretraining: same cohort, from scratch (pilot); no pretraining: other cohorts, from scratch (pilot); rare states: from scratch vs pretrained; deconvolution with the same pairs: composition; deconvolution with the same pairs: disease signal |
+| Appendix | scSemiProfiler without its output threshold (sensitivity; modified read-out, not the official output) |
+
+## Rebuild
+
+Everything is CPU only; no GPU or checkpoint is touched.
+
+```bash
+# prerequisites (plan folder plans/paired_finetune_intradataset_20260927/, CPU):
+#   scsemi/tools/representative_dependence.py      -> scsemi/analysis/representative_dependence_20261002/
+#   tools/contrast_extra.py (+ tools/contrast_extra_tables.py) -> scores_contrast/   (Spearman, ceiling, limma, CAT, RBO)
+#   tools/build_tpm_zeroshot.py + tools/score_tpm_zeroshot.sh -> scores/<ds>/zeroshot_tpm_20261002/pooled/
+#     (the length-corrected zero-shot arm needs one GPU sampling step: tools/sample_lockwait.sh, coordinator only)
+cd /home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/presentations/2026-09-30--paper-summary--r00
+# shared helpers, not run directly: style.py, sec03_common.py, scsemi_results.py, fold_sizes.py
+for f in sec01_overview sec02_backbone sec03_width_loss sec03_width_bench sec03_panel \
+         sec04_metrics_overview sec04_sle_benchmark sec05_condition_space sec05_cells_offmanifold \
+         sec05_behcet_direction sec05_shift_cohorts sec06_finetune_method sec07_ft_modes \
+         sec07_k_aug sec07_rank_sweep sec08_pairs_needed sec08_cross_cohort sec09_paired_validation sec09_extra_metrics \
+         sec09_time_to_apply sec09_scsemi_representative sec09_cat_curves sec10_paired_subset_cohorts \
+         secA_scsemi_readout sec07_no_pretraining sec08_cross_scratch sec09_same_pairs \
+         secS_scratch_rare_states secS_deconvolution fig3_time_days sec10_other_tissues figure_plan; do   # keep this order (sec10 reads secS; figure_plan last)
+  python3 "tools/$f.py" || break
+done
+python3 tools/restructure_figure_order_20261006.py   # order the slides by data/figure_plan.json
+python3 tools/sync_sec09_text.py      # copy every generated slide text (sec09_index / text_source) into the spec
+python3 generate_deck.py              # -> PBISC-Diffusion_paper_summary_20260930.pptx
+soffice --headless --convert-to pdf --outdir render PBISC-Diffusion_paper_summary_20260930.pptx
+pdftoppm -jpeg -r 70 render/PBISC-Diffusion_paper_summary_20260930.pdf render/slide
+cp render/PBISC-Diffusion_paper_summary_20260930.pdf .
+```
+
+`generate_deck.py` stops with an error if a figure is missing, a title does not fit on one line, a footnote needs
+more than 2 lines (1 line when the slide has references), the reference line needs more than 1 line, or any text
+contains Korean or placeholder words. `tools/restructure_20261002.py` is the one-off script that produced the current
+slide order and the references (it refuses to run twice).
+
+## Notes
+
+- **No "all" label (2026-10-02).** The 5-fold arm trains each fold model on the other four folds minus 7–8
+  inner-validation donors; its test fold is never trained on (RUN.json `test_rows_never_loaded_for_training`). Figures
+  label it "5-fold CV (other 4 folds)" and learning-curve ticks show the training-pair count per fold (78–79 ImMicro,
+  58–59 sepsis), read by `tools/fold_sizes.py`.
+- **Learning curve x = 0 uses no pairs (2026-10-02).** It shows the zero-shot model fed the real bulk as measured
+  and after exon-length correction (counts / Ensembl union-exon kb, then the same fixed-depth arithmetic;
+  `plans/.../tools/build_tpm_zeroshot.py`, scored in `scores/<ds>/zeroshot_tpm_20261002/pooled`). The old x = 0
+  calibration point, whose alpha used every training pair of the fold, is gone; calibration appears only fitted on the
+  same n pairs as the LoRA point (squares; `cal_budget_20261001`).
+- **Disease metrics (2026-10-02).** Next to Pearson logFC r and the Welch top-50 DEG overlap: Spearman r, the
+  split-half ceiling (observed vs observed, Spearman–Brown), limma-voom top-50 DEG, each arm's chance overlap, and
+  CAT curves with the k / 2G chance line (`plans/.../scores_contrast/`, tools/contrast_extra.py; every Pearson r and
+  Welch count reproduced exactly). Repeated samples of one patient (Imperial, Giroux, Ji) count as separate samples in
+  both DE tests, as in the scorer (footnote).
+- Section 9 text is generated: `tools/sec09_paired_validation.py`, `sec09_extra_metrics.py`, `sec07_rank_sweep.py`,
+  `sec09_time_to_apply.py`, `sec09_scsemi_representative.py`, `sec09_cat_curves.py` write slide text JSON; sync copies
+  it. Hand-written: the keywords of the two "Pairs needed" slides (section 8) and of every slide in sections 1–7 and 10.
+- Giroux's disease metrics exclude 196 sex-linked genes (severity is confounded with sex), from
+  `plans/.../scores_nosex/`. The calibration comparator is "Zero-shot + calibration" (user decision 2026-10-01).
+- Story rule (user decision 2026-09-30): no PBISC-VAE and no reference calibration (refcal). The exon-length-corrected
+  input is used only as the no-pair x = 0 point of the learning curve (user decision 2026-10-02).
+  `tools/sec04_immicro.py` (refcal vs scSemiProfiler on ImMicro) is kept but is not part of the deck or the rebuild.
+- scSemiProfiler 5-fold and equal-budget results are picked up automatically when they appear (`tools/scsemi_results.py`;
+  arm `scsemi__raw` in `scores/<cohort>/scsemi_cv_20261002/pooled*/`, `scsemi_n{5,10,20,40}__raw` in
+  `scores/<cohort>/scsemi_budget_20261002/pooled*/`; same donors and reference cells as the main pooled call only).
+  Then rerun sections 8 and 9, sync, recheck the section 8 keywords and rebuild.
+- Earlier versions: `_archive/2026-09-30_before_no_vae_refcal/`, `_archive/2026-10-01_before_extra_metrics/`,
+  `_archive/2026-10-02_before_time_to_apply/`, `_archive/2026-10-02_before_restructure/` (pptx, PDF, spec; their
+  figures are in the paper archive). `DRAFT_1150_*`: the copy shown at the 2026-09-30 12:00 meeting.
+
+- Added 2026-10-03 (user decision: the cross-cohort experiment is reported at K = 1000 only and is the figure that explains
+  why PBISC fits LoRA on a few pairs of each cohort): slides 25-26 in section 8, `tools/sec08_cross_cohort.py`.
+  - Slide 25 "Pairs from other cohorts do not transfer": per held-out cohort, composition L1 / coverage / disease logFC r of
+    zero-shot, calibration fitted on the other cohorts, LoRA r16 / r64 / full fine-tune trained on the other cohorts
+    (+ Liu 2021) at step saturation, and LoRA on the cohort's own pairs (5-fold pooled).
+  - Slide 26 "Longer training on other cohorts" (": gap stays" since 2026-10-06): the same metrics at every 20k-step
+    milestone.
+  - Source plan: `plans/paired_crosscohort_20261002/` (README; same-reference and same-sample checks in the script).
+    Agenda item 8 now reads "How many pairs, and do other cohorts' pairs transfer". The deck has 37 slides.
+    Pre-update copy: `_archive/2026-10-03_before_cross_cohort/`.
+
+## Sources
+
+Final model: `runs/f01_6k_1x_seed20260627` (epoch 16, EMA). Scores and receipts come from
+`plans/utility_studies_final6k_20260925/`, `plans/paired_finetune_intradataset_20260927/`,
+`plans/immicro_scsemi_comparison_20260927/`, `plans/floor01_campaign_20260923/`,
+`analysis/scsemi_benchmark_diffusion_20260913/` and the run folders named in each `data/sec*.csv`. Reference PDFs:
+`/home/sj_server_1/PBISC/paper-release-20260928/reference-papers/` (INDEX.md). Manuscript:
+`/home/sj_server_1/PBISC/paper-release-20260928/manuscript/`.
+- 2026-10-04 appendix (user decision: supplementary, appendix only, because it changes scSemiProfiler's official
+  output). Slide 38 is the divider "Appendix" and slide 39 is "Appendix: scSemiProfiler without output threshold"
+  (`tools/secA_scsemi_readout.py`, `data/secA_scsemi_readout{.csv,_slide_text.json}`).
+  - Inputs: `plans/paired_finetune_intradataset_20260927/scsemi/analysis/output_variants_20261003/`. The script checks
+    that its LoRA-5 and official scSemiProfiler values equal the section 9 slide data draw by draw.
+  - Result: without upstream's output threshold, scSemiProfiler's coverage rises in 5/5 cohorts, and 23/23 LoRA-5
+    comparisons (coverage, precision, composition, logFC r, DEG) keep their winner.
+  - The outline keeps its 10 sections (the builder requires 10). The pre-change copy is in
+    `_archive/2026-10-04_before_appendix/`. The deck has 39 slides.
+- 2026-10-05 update (user request: put the results so far into the deck). Pre-change copy: `_archive/2026-10-05_before_scratch_cv/`.
+  - New slide 23, "No pretraining: same cohort, from scratch" (`tools/sec07_no_pretraining.py`). Pilot on sepsis fold 0
+    (17 donors). Reads `fold0_scratch_20261005/` and the runs' RUN.json. The donor composition r is recomputed from
+    per_donor.csv and reproduces ablation_extra_donor.csv.
+  - New slide 28, "No pretraining: other cohorts, from scratch" (`tools/sec08_cross_scratch.py`). Held-out sepsis pilot.
+    Every arm is read from `paired_crosscohort_20261002/scores_masked/`: genes no training cohort measured are zeroed
+    in all arms, because the unmasked scratch arm puts 91–94 % of its UMI there (read from the generated_masked logs
+    since 2026-10-06).
+  - New slide 38, "Same training pairs: LoRA vs scSemiProfiler" (`tools/sec09_same_pairs.py`). scSemiProfiler 5-fold
+    CV with LoRA's own training pairs, plus GPU hours per fold from the receipts on either host. Cohorts that are still
+    running show "in progress"; rerun the script as they finish.
+  - These two slides now draw the scSemiProfiler CV / budget runs automatically:
+    - Slide 24, "Pairs needed" (diamonds; ImMicro so far): keyword 2 reworded, footnote extended.
+    - Slide 37, "5-fold CV" (yellow bar; ImMicro and Ji so far).
+  - `tools/sec09_paired_validation.py` reads the scSemiProfiler CV arm's Spearman / limma values from
+    `scores_contrast/<cohort>/scsemi_cv_20261002/pooled/` (contrast_extra.py --design scored). Pearson r and the Welch
+    DEG must match the scorer; the values stay blank until computed. Cohorts not yet run say "in progress".
+  - Agenda item 7 now reads "LoRA: compute, rank, K; training from scratch". The deck has 42 slides.
+  - New helper: `tools/grouped_bars.py`.
+- 2026-10-06 update (user request: check that the cross-cohort LoRA / full FT, cross-cohort from-scratch and
+  same-cohort from-scratch (+ recipes) results, with benchmarks and training times, and the scratch-vs-pretrained rare
+  states are in the deck; user decision: the from-scratch results go to the paper's Supplementary and the Discussion
+  presents them as a route to other tissues). Pre-change copy: `_archive/2026-10-06_before_supplementary/`.
+  - **Supplementary block (new divider, slides 40-43).**
+    - "No pretraining: same cohort, from scratch" and "No pretraining: other cohorts, from scratch" moved here from
+      sections 7 and 8. Their content is unchanged; slide 42's notes now name the cross-cohort slides by title.
+    - New "Rare states: from scratch vs pretrained" (`tools/secS_scratch_rare_states.py`).
+      - It reads the rare-state screen (`plans/rare_states_screen_20261004/out/sepsis_fold0_*` and
+        `sepsis_loco_masked_*`).
+      - The cross-cohort arms use the masked scoring (46 never-measured genes zeroed in every arm).
+      - The donor bootstrap reproduces the verified analysis in `out/scratch_vs_lora_20261005/REPORT.md`.
+  - **Section 10 "Next steps".**
+    - New "Next: other tissues from a few dozen pairs" (`tools/sec10_other_tissues.py`): a comparison table, pretrained
+      + LoRA vs from scratch.
+    - Every value is read from the deck's own source tables. The say text is the user's Discussion framing.
+  - **Slide 26 "Longer training on other cohorts".** The first keyword is now training minutes to saturation (RUN.json
+    wall seconds), with per-cohort minutes and steps in the notes (`tools/sec08_cross_cohort.py`).
+  - **Slide 34 "5 donors: time to apply".**
+    - The pretraining hours are read from `data/sec06_ft_method.csv`, no longer hard-coded.
+    - The note names the 'Fine-tuning with paired bulk + sc' slide by title. Its old "slide 15" pointed at the wrong
+      slide.
+  - Agenda items 7 and 10 updated. Spec change: `tools/restructure_20261006.py` (refuses to run twice). The deck has 45
+    slides; no Korean text.
+  - **Review pass (same day).** An independent 5-agent review of the changed slides confirmed 39 issues; all are
+    fixed. Pre-review copies of the touched scripts and the spec: `_archive/2026-10-06_before_supplementary/*.pre_review`.
+    - Slide 39: pretraining now reads 33.1 M train cells (`train_split_cells`; the 41.4 M corpus includes val/test).
+      The rare-state row follows the bootstrap verdicts on both sides (closer on ISG-high T / on DN2 B; labels and
+      states both arms miss by > 4-fold are not counted). "No corpus" became "come close": the scratch runs still took
+      the panel, value scaler and z-score constants from the PBMC checkpoint (RUN.json `kept_from_checkpoint`, now in
+      the notes). The say text carries the one-fold pilot qualifier; the footnote names both pilots.
+    - Slide 43: "no significant difference on average" (not "no difference"). Keyword 2 is the cycling T/NK count
+      (8 of 8 comparisons). The notes add the ISG-high T split by panel, the scratch-closer states (DN2 B, MS1), the
+      52-call tally, the fold-0 caveat (fold 0 is the own-pair LoRA's best fold for C1Q-high and cycling T/NK,
+      computed from `folds.csv`), label states unthinned, and "8 of 13 states shown".
+    - Slide 26: title "...: gap stays" and "drifts in Sepsis and Ji" are computed from the milestones (gap = no milestone
+      reaches the own-cohort coverage; drift = full fine-tune L1 up by >= 0.05). Minutes are in
+      `data/sec08_cross_cohort.csv` (wall_seconds, minutes, run_receipt) and in the numbers list.
+    - Slide 42: UMI share (94 %, 91–94 % across milestones), the observed share (0.014 %, from `cells_panel6k.npz`)
+      and both arms' stopping settings are read from the receipts; the "none" label sits inside the axes.
+    - Slide 34: keywords use the cohort means of the figure (17–59×, 57 min–4.1 h); the per-draw range 11–63× is in the
+      notes. The right panel shares the bars' rows and has a title.
+    - Talk text and pointers: slides 23, 25, 26, 45 (say), 22 (recipes description), 32 and 36 (slide pointers by
+      title), slide 1 subtitle "updated 2026-10-06".
+- 2026-10-08 update (user request: put the BayesPrism, CIBERSORTx and scSemiProfiler 5-fold results into the deck).
+  Pre-change copy: `_archive/2026-10-08_before_deconvolution/`.
+  - **Two new Supplementary slides** (`tools/secS_deconvolution.py`, `data/secS_deconvolution{.csv,_slide_text.json}`):
+    "Deconvolution with the same pairs: composition" and "...: disease signal".
+    - Source: `plans/bayesprism_paired_20261006/` (README, `results/RESULTS_full.md`, `RESULTS_cibersortx.md`).
+    - BayesPrism (recommended genes) and official CIBERSORTx (Fractions S-mode + HiRes) use the same paired donors
+      as the reference and the test samples' real bulk as the mixture, with the benchmark's own truth and scorer.
+    - Keywords and notes are computed from the tables.
+  - **Supplementary plan.** The planned "Bulk deconvolution methods" row is replaced by the two Ready slides.
+    `tools/figure_plan.py` now takes an optional per-slide note (`supplementary.notes`) and drops "Planned:" when
+    nothing is planned.
+  - **New scSemiProfiler results picked up (server_1):**
+    - Sepsis and Giroux 5-fold CV. Giroux was finalized on server_1 from sha-checked copies of its fold NPZs, and
+      rescored without sex-linked genes in `scores_nosex/giroux2022/scsemi_cv_20261002/pooled`.
+    - Sepsis equal-budget n = 5/10/20/40.
+    - Imperial 5-fold: finished on server_2 at 13:45 and was finalized on server_1 from sha-checked fold copies.
+      scSemiProfiler 5-fold now covers all five cohorts.
+    - Spearman/limma extras for sepsis, Giroux and Imperial; realism extras for all five cohorts.
+  - **Text changes from the new results:**
+    - "Same training pairs": keywords are data-driven. LoRA leads on coverage and precision in 5/5 cohorts;
+      composition 4/5 and logFC r 3/4, where scSemiProfiler leads in Giroux. The GPU-time ratio is 19–62×.
+    - sec09_same_pairs.py now reads the sepsis fold-0 pilot queue tag (`sep_f0`). For Giroux it quotes DEG from
+      scores_nosex, and for older pooled LoRA calls from pooled_deg.
+    - "5-fold CV" keyword: logFC r 3/4.
+    - "Pairs needed" (hand-written): scSemiProfiler coverage 0.04 / 0.10. On sepsis disease logFC r,
+      scSemiProfiler at 5–40 pairs (0.52–0.58) is above LoRA (0.29–0.48), while its top-50 DEG stays at 2–6.
+- 2026-10-08, new figure split (user decision; supersedes the 2026-10-06 split; pre-change copy
+  `_archive/2026-10-08_before_new_figure_plan/`).
+  - Fig. 2: LoRA fine-tuning with no disease metrics.
+  - Fig. 3: PBISC vs scSemiProfiler on healthy donors, with time as a main result.
+  - Fig. 4: disease benchmarks on the paired cohorts (PBISC versions, scSemiProfiler, and BayesPrism / CIBERSORTx
+    labelled as cell-type deconvolution; the deconvolution slides moved here from the Supplementary).
+  - Fig. 5: rare-cell fractions and detection.
+  - Fig. 6: new findings from the bulk-only sepsis and COVID samples (K = 1,000 and 4,000).
+  - The cross-cohort transfer slides moved to the Supplementary; their placement is open.
+  - New slide "Time: days for scSemiProfiler, minutes for LoRA" (`tools/fig3_time_days.py`, Fig. 3d).
+    - Inputs: GPU hours per 5-fold experiment from the receipts (`data/sec09_same_pairs.csv`, checked against
+      the scSemiProfiler event logs on both hosts), the 5-donor draws (`data/sec09_time_to_apply.csv`) and the
+      one-time pretraining (`data/sec06_ft_method.csv`).
+    - Results: ImMicro 24 h vs 1.2 h, sepsis 38 h vs 37 min. All runs: scSemiProfiler 184 h (7.7 days) vs PBISC
+      22 h including pretraining.
+    - The equal-budget scSemiProfiler runs are excluded: they reuse the 5-fold run's representative models.
+  - Panel status: Build / Run marks new slides or analyses still to do; Adapt marks slides to re-scope (healthy-only
+    Fig. 2-3, disease columns into Fig. 4).

paper-releases/pbisc-diffusion/updates/20261008T060837Z-deck-new-figure-split-time/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/tools/fig3_time_days.py

@@ -0,0 +1,194 @@
+#!/usr/bin/env python3
+"""Fig. 3 time slide of the 2026-09-30 paper-summary deck: what one experiment costs, scSemiProfiler vs PBISC LoRA
+(2026-10-08; user request: time is a main result, scSemiProfiler takes days per experiment, show it in numbers).
+
+CPU only, plotting only. Same GB10 GPU type for both methods, every step alone on the GPU.
+  5-fold CV per cohort   data/sec09_same_pairs.csv (tools/sec09_same_pairs.py): GPU hours per fold model, mean over
+                         the 5 folds (LoRA = fine-tune + sampling 1,500 cells per test sample; scSemiProfiler = every
+                         representative's two pretraining stages + one model per test sample, from its receipts).
+                         One experiment = the 5 fold models (mean x 5).
+  5-donor draws          data/sec09_time_to_apply.csv (tools/sec09_time_to_apply.py), minutes per draw, 5 draws.
+  PBISC pretraining      data/sec06_ft_method.csv pretrain_wall_hours: done once, reused by every cohort.
+  scSemiProfiler running time (wall clock, CPU steps included): scsemi/work/<cohort>/cv/fold<k>/events.jsonl on the
+                         host that ran each fold; gaps over 30 min between events (waiting for the GPU lock) excluded.
+Not counted on either side: data preparation before the runs, scoring. The equal-pair-budget scSemiProfiler runs are
+left out: they reuse the representative models of the 5-fold run, so their receipts understate a fresh run.
+Outputs: figures/fig3_time_days.{png,pdf}, data/fig3_time_days.csv, data/fig3_time_days_slide_text.json
+"""
+from __future__ import annotations
+
+import json
+import os
+import subprocess
+import sys
+from pathlib import Path
+
+os.environ["CUDA_VISIBLE_DEVICES"] = ""  # plotting only
+
+import numpy as np
+import pandas as pd
+
+DECK = Path(__file__).resolve().parents[1]
+sys.path.insert(0, str(DECK / "tools"))
+import style as ST  # noqa: E402
+import matplotlib.pyplot as plt  # noqa: E402
+
+T = Path("/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927")
+FIGDIR, OUTDATA = DECK / "figures", DECK / "data"
+ORDER = ["immicro", "ye2026_sepsis", "imperial_eskd", "giroux2022", "ji2025_chili"]
+NAME = {"immicro": "ImMicro (healthy)", "ye2026_sepsis": "Sepsis", "imperial_eskd": "Imperial", "giroux2022": "Giroux",
+        "ji2025_chili": "Ji (whole blood)"}
+COL = {"scsemi": ST.ARM["scsemi"], "lora": ST.ARM["lora"], "pretrain": "#F2B48C"}
+WAIT_S = 1800
+
+COLLECT = r'''
+import glob, json, os, sys
+from datetime import datetime
+T = sys.argv[1]; WAIT = float(sys.argv[2]); out = {}
+for p in glob.glob(os.path.join(T, "scsemi", "work", "*", "cv", "fold*", "events.jsonl")):
+    ev = []
+    for line in open(p):
+        try: ev.append(json.loads(line))
+        except Exception: pass
+    if not ev or ev[-1]["event"] != "infer_call_done": continue
+    ts = [datetime.fromisoformat(e["at"]) for e in ev]
+    gaps = [(b - a).total_seconds() for a, b in zip(ts, ts[1:])]
+    rel = os.path.relpath(os.path.dirname(p), os.path.join(T, "scsemi", "work"))
+    out[rel] = {"active_s": sum(g for g in gaps if g <= WAIT), "waits_s": sum(g for g in gaps if g > WAIT),
+                "gpu_s": sum(e.get("elapsed_seconds", 0) for e in ev if e["event"] in ("pretrain_complete", "target_complete"))}
+print(json.dumps(out))
+'''
+
+
+def collect(host: str) -> dict:
+    cmd = ["python3", "-", str(T), str(WAIT_S)]
+    if host != "server_1":
+        cmd = ["ssh", "spark-0110"] + cmd
+    return json.loads(subprocess.run(cmd, input=COLLECT, capture_output=True, text=True, check=True).stdout)
+
+
+def fmt_h(h: float) -> str:
+    if h >= 20:
+        d = h / 24
+        return f"{h:.0f} h ({d:.1f} day{'' if round(d, 1) == 1.0 else 's'})"
+    return f"{h:.1f} h" if h >= 1 else f"{h * 60:.0f} min"
+
+
+def main() -> int:
+    sp = pd.read_csv(OUTDATA / "sec09_same_pairs.csv").set_index(["dataset", "arm"])
+    ta = pd.read_csv(OUTDATA / "sec09_time_to_apply.csv").set_index(["cohort", "method"])
+    ft = pd.read_csv(OUTDATA / "sec06_ft_method.csv").set_index("key")
+    pre_h = float(ft.loc["pretrain_wall_hours", "value"])
+    runs = {}
+    for h in ("server_1", "server_2"):
+        for k, v in collect(h).items():
+            if k in runs:
+                raise SystemExit(f"{k}: finished on both hosts")
+            runs[k] = v
+    rows = []
+    for ds in ORDER:
+        folds = [runs.get(f"{ds}/cv/fold{k}") for k in range(5)]
+        if any(f is None for f in folds):
+            raise SystemExit(f"{ds}: scSemiProfiler 5-fold not complete on either host")
+        gpu_ev = sum(f["gpu_s"] for f in folds) / 3600
+        sc_cv = 5 * float(sp.loc[(ds, "scsemi"), "hours"])
+        if abs(gpu_ev - sc_cv) > 0.02 * sc_cv:      # same receipts, two read-outs
+            raise SystemExit(f"{ds}: event GPU time {gpu_ev:.2f} h differs from sec09_same_pairs {sc_cv:.2f} h")
+        rows.append({"cohort": ds, "cv_scsemi_gpu_h": sc_cv, "cv_lora_gpu_h": 5 * float(sp.loc[(ds, "lora"), "hours"]),
+                     "cv_scsemi_running_h": sum(f["active_s"] for f in folds) / 3600,
+                     "cv_scsemi_gpu_waits_excluded_h": sum(f["waits_s"] for f in folds) / 3600,
+                     "draws_scsemi_gpu_h": float(ta.loc[(ds, "scsemi"), "total_min_mean"]) * 5 / 60,
+                     "draws_lora_gpu_h": float(ta.loc[(ds, "lora"), "total_min_mean"]) * 5 / 60})
+    tab = pd.DataFrame(rows)
+    tab["cv_ratio"] = tab.cv_scsemi_gpu_h / tab.cv_lora_gpu_h
+    tot_sc = tab.cv_scsemi_gpu_h.sum() + tab.draws_scsemi_gpu_h.sum()
+    tot_lo = tab.cv_lora_gpu_h.sum() + tab.draws_lora_gpu_h.sum()
+    tab.to_csv(OUTDATA / "fig3_time_days.csv", index=False, float_format="%.4g")
+
+    # ------------------------------------------------------------------ figure
+    fig, (a1, a2) = plt.subplots(1, 2, figsize=ST.FULL, gridspec_kw={"width_ratios": [1.65, 1], "wspace": 0.45})
+    y = np.arange(len(ORDER))
+    h = 0.36
+    for i, r in tab.iterrows():
+        for k, (arm, v) in enumerate((("scsemi", r.cv_scsemi_gpu_h), ("lora", r.cv_lora_gpu_h))):
+            yy = y[i] + (k - 0.5) * h
+            a1.barh(yy, v, h * 0.9, color=COL[arm], edgecolor="none")
+            a1.text(v * 1.12, yy, fmt_h(v), va="center", fontsize=ST.FS - 4, color=ST.SUBINK)
+    for d, lab in ((24, "1 day"), (48, "2 days")):
+        a1.axvline(d, color="#BBBBBB", linestyle=":", linewidth=1, zorder=0)
+        a1.text(d, -0.62, lab, ha="center", va="center", fontsize=ST.FS - 5, color=ST.SUBINK)
+    a1.set_xscale("log")
+    a1.set_xlim(0.1, 400)
+    a1.set_xticks([0.1, 1, 10, 100], ["0.1", "1", "10", "100"])
+    a1.minorticks_off()
+    a1.set_yticks(y, [NAME[d] for d in ORDER], fontsize=ST.FS - 2)
+    a1.set_ylim(len(ORDER) - 0.5, -0.75)
+    a1.set_xlabel("GPU hours for one 5-fold experiment (log)")
+    a1.set_title("One cohort, 5-fold CV (5 fold models)", fontsize=ST.FS - 1, loc="left")
+    # right: everything run for this paper (5-fold + 5-donor draws, 5 cohorts)
+    bars = [("scSemiProfiler", [(tot_sc, COL["scsemi"])]),
+            ("PBISC", [(tot_lo, COL["lora"]), (pre_h, COL["pretrain"])])]
+    for i, (lab, parts) in enumerate(bars):
+        left = 0.0
+        for v, c in parts:
+            a2.barh(i, v, 0.55, left=left, color=c, edgecolor="none")
+            left += v
+        a2.text(left + 3, i, fmt_h(left), va="center", fontsize=ST.FS - 4, color=ST.SUBINK)
+    a2.text(tot_lo + pre_h + 3, 1.22, f"LoRA {tot_lo:.1f} h + pretraining {pre_h:.1f} h (once)", ha="left",
+            va="center", fontsize=ST.FS - 5, color=ST.SUBINK)
+    a2.set_yticks([0, 1], [b[0] for b in bars], fontsize=ST.FS - 2)
+    a2.set_ylim(1.6, -0.6)
+    a2.set_xlim(0, tot_sc * 1.45)
+    a2.set_xlabel("GPU hours, all runs (linear)")
+    a2.set_title("All runs for this paper", fontsize=ST.FS - 1, loc="left")
+    for ax in (a1, a2):
+        ax.grid(axis="x", color="#E3E3E3", linewidth=0.8)
+        ax.set_axisbelow(True)
+        ax.tick_params(axis="y", length=0)
+    handles = [plt.Rectangle((0, 0), 1, 1, color=COL[k]) for k in ("scsemi", "lora", "pretrain")]
+    fig.legend(handles, ["scSemiProfiler", "PBISC LoRA (fine-tune + sampling)", "PBISC pretraining, once"],
+               loc="lower center", ncol=3, bbox_to_anchor=(0.5, -0.07), fontsize=ST.FS - 3)
+    ST.save(fig, "fig3_time_days", FIGDIR)
+
+    # ------------------------------------------------------------------ text
+    im = tab.set_index("cohort").loc["immicro"]
+    worst = tab.loc[tab.cv_scsemi_gpu_h.idxmax()]
+    kw = [f"Healthy 5-fold: {im.cv_scsemi_gpu_h / 24:.1f} GPU-day vs {im.cv_lora_gpu_h:.1f} h",
+          f"Up to {worst.cv_scsemi_gpu_h / 24:.1f} days per cohort ({NAME[worst.cohort]})",
+          f"All runs: {tot_sc / 24:.1f} days vs {tot_lo + pre_h:.0f} h with pretraining"]
+    run_txt = "; ".join(f"{NAME[r.cohort]} {r.cv_scsemi_running_h:.0f} h" for r in tab.itertuples())
+    full = ("GPU time on one GB10 GPU with every step alone on the GPU, from each run's receipts. One 5-fold experiment "
+            "= the five fold models, every usable sample tested once by the model that never trained on it. "
+            "scSemiProfiler = pretraining of every representative (two stages) + one model per test sample "
+            "(upstream c2aea02, default settings); PBISC LoRA = fine-tune + sampling 1,500 cells per test sample. "
+            f"Ratio per cohort: {tab.cv_ratio.min():.0f}-{tab.cv_ratio.max():.0f}x. scSemiProfiler running time "
+            f"including its CPU steps (waits for the GPU excluded): {run_txt}. All runs = 5-fold + the 25 five-donor "
+            f"draws: scSemiProfiler {tot_sc:.0f} h, LoRA {tot_lo:.1f} h; PBISC's pretraining ({pre_h:.1f} h, 33.1 M "
+            "cells) is done once and reused by every cohort. Not counted on either side: data preparation and scoring. "
+            "The equal-pair-budget scSemiProfiler runs are not included (they reuse the 5-fold run's representative "
+            "models). Sources: data/sec09_same_pairs.csv, data/sec09_time_to_apply.csv, data/sec06_ft_method.csv, "
+            "scsemi/work/<cohort>/cv/fold<k>/events.jsonl.").replace("-", "–")
+    numbers = [{"value": "; ".join(f"{NAME[r.cohort]}: 5-fold scSemiProfiler {r.cv_scsemi_gpu_h:.1f} GPU h "
+                                    f"({r.cv_scsemi_gpu_h / 24:.2f} days, running {r.cv_scsemi_running_h:.1f} h) vs LoRA "
+                                    f"{r.cv_lora_gpu_h:.2f} h ({r.cv_ratio:.0f}x); 5-donor draws {r.draws_scsemi_gpu_h:.1f}"
+                                    f" h vs {r.draws_lora_gpu_h:.2f} h" for r in tab.itertuples()),
+                "source": "data/fig3_time_days.csv"},
+               {"value": f"all runs: scSemiProfiler {tot_sc:.1f} GPU h ({tot_sc / 24:.2f} days); PBISC LoRA {tot_lo:.2f} h"
+                         f" + pretraining {pre_h:.1f} h once", "source": "data/fig3_time_days.csv"}]
+    slides = [{"title": "Time: days for scSemiProfiler, minutes for LoRA", "layout": "full",
+               "figures": [str(FIGDIR / "fig3_time_days.png")], "keywords": kw,
+               "footnote_short": "One GB10 GPU, job alone on it. GPU time from receipts; waits for the GPU excluded.",
+               "footnote": full, "numbers": numbers}]
+    if len(slides[0]["footnote_short"]) > 110:
+        raise SystemExit("short footnote too long")
+    out = {"section": "fig3-time", "scripts": [str(Path(__file__).resolve())],
+           "generated_from": "tools/fig3_time_days.py (rerun to refresh; do not edit by hand)", "slides": slides}
+    (OUTDATA / "fig3_time_days_slide_text.json").write_text(json.dumps(out, indent=1, ensure_ascii=False))
+    print("\n".join(kw))
+    print(tab.round(2).to_string(index=False))
+    print(f"all runs: scSemi {tot_sc:.1f} h, LoRA {tot_lo:.2f} h, pretraining {pre_h:.1f} h")
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())