Archive 4 checkpoints (1-4)

paper-releases/pbisc-diffusion/updates/20261003T174003Z-scsemi-readout-sensitivity/diffusion/dataset/server_1/model8/plans/paired_finetune_intradataset_20260927/scores/giroux2022/fewshot_n5/draw2_scsemi5nt/README.md

@@ -0,0 +1,73 @@
+# score_ft.py output: giroux2022, 12 scored donors
+
+Arms: scsemi5nt.  Every arm is scored on the same donors against the same observed reference.
+
+## Inputs and fixed contracts
+- Observed reference: `/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scoring_cache/giroux2022/reference.npz` (sha dd0152f7bde4), built from datadir:/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data/giroux2022;
+  observed cells sum exactly to the features pseudobulk of every donor. Observed labels: /home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scoring_cache/giroux2022/observed_labels/labels/labels.parquet.
+- Generated cells: `generated_counts` read out once with Poisson(seed 20260916) over the **whole NPZ in row order**
+  (`poisson_readout.poisson_integerise`), then the scored donors' rows. These integer counts are used for labels, embedding,
+  depth and the contrast. Pseudobulk uses the summed `generated_counts` (rates), as `score_generated.py`.
+- Labels: frozen f01common scANVI (`predict_scanvi.py`, CPU, zero query epochs), six-type gate ['T cell', 'B cell', 'Monocyte', 'NK cell', 'DC', 'Other']; query genes =
+  the 1,006-gene annotation list minus genes the dataset's sc assay did not measure in every sample (1006 genes).
+- Scoring genes: 6k panel genes measured by both assays in every sample of the dataset (5563 genes).
+
+## Definitions
+1. **Composition**: fraction of each donor's cells per six-type label (all generated cells; all observed QC cells).
+   `comp_l1` = sum over the six labels of |generated - observed| (immicro_common.l1). `comp_l1_obs_loo_null` = L1 of the
+   donor's observed composition to the mean of the *other scored donors'* observed compositions. `composition_top1` =
+   fraction of donors whose generated composition is closest (L1) to their own observed composition.
+2. **Pseudobulk**: `pb_r` = Pearson r of CPM-log1p(summed generated_counts) vs CPM-log1p(observed pseudobulk) on the scoring
+   genes (evaluate_realbulk.cpm_log1p / paired_r). `pb_top1` = the donor's generated pseudobulk is closest (Pearson) to its own
+   observed pseudobulk among the 12 scored donors (immicro_common.retrieval = evaluate_realbulk.top1 arithmetic).
+3. **Precision / coverage / density** (stage_b_embed.py functions): per-cell CP10K over the scoring genes + log1p; PCA-50
+   (randomized, random_state 20260925) and k = 10 real-real k-NN balls fitted on **R1 of the scored donors only**.
+   Reference rule (independent of fold and arm): per donor, cells in canonical order (library_id+barcode), one
+   permutation from `default_rng([20260925, uint64(sha256(donor)[:8])])`; R1 = first n, R2 = next n,
+   n = min(150, n_cells // 2) (1800 R1 cells here). Generated query: per donor n cells (the donor's R1 size),
+   without replacement, `default_rng([20260926, same key])`. Precision = fraction of generated cells inside any R1 ball
+   (Kynkaanniemi 2019); coverage = fraction of R1 cells whose ball holds a generated cell; density = Naeem 2020.
+   Per-donor precision = that donor's generated cells; per-donor coverage = that donor's R1 cells (the pooled values are
+   their n-weighted means; plain means when every donor has 150 reference cells). R2 (held-out real cells of the same donors) is the
+   perfect-generator control: precision 0.9989, coverage 1.0000.
+   **Depth matched**: stage_b_embed depth matching run pairwise (R1 vs this arm): per donor and within-donor rank of detected
+   genes, both cells are thinned by UMI sampling without replacement to the smaller detected-gene count; PCA kept from
+   un-thinned R1, radii recomputed on thinned R1; fresh rng(20260929) per arm. R2 control: precision
+   0.9994, coverage 1.0000.
+4. **Monocyte %** = 100 x Monocyte fraction (all cells) per donor; **median UMI per cell** = median over cells of the
+   Poisson-read (generated) / observed counts summed over the scoring genes (per donor, and pooled).
+5. **Group contrast** (T, B, Monocyte, NK): case groups ['severe'] (n = 6) vs control groups ['mild'] (n = 6); rule: explicit --case-groups / --control-groups; groups from features groups.
+   Genes per type = scoring genes detected (count > 0) in >= 1 % of the scored donors' observed
+   cells of that type. Primary statistic `mean_cp10k`: per-cell CP10K over the scoring genes; per donor the mean over its
+   cells of the type (donors with >= 1 such cell); group mean = mean over the group's donors (donor-balanced);
+   logFC = log2(case mean + 1) - log2(control mean + 1). Secondary `pseudobulk_log1p_cpm`: the SLE benchmark evaluator's
+   statistic (donor x type summed counts, log1p CPM over the scoring genes, mean case - mean control;
+   evaluate_generated_vs_truth.group_log_fold_change). Both are computed on the generated cells (their frozen labels) and on
+   the observed cells; `r_generated` = Pearson r of the two logFC vectors. Broadcast baseline `r_broadcast`: the same
+   observed vector against the logFC of the input bulk (condition NPZ `bulk_counts` of the arm, scoring genes, per-donor
+   CP10K resp. log1p CPM, group means). `partial_r_generated_given_broadcast`: r of the OLS residuals of observed and
+   generated logFC on the bulk logFC (residual_specificity.py).
+6. **Top-50 DEG overlap** (contrast.csv rows `statistic = welch_top50_deg`; summary `deg_top50`): the SLE benchmark's score (`total_top50_overlap` of
+   `bulk2sc-paper/scripts/evaluate_generated_vs_truth.py`, Welch backend: the "top-50 DEG 81 / 84" of the final model), run
+   with the evaluator's own functions on the same donor x type pseudobulks as the contrast. Per cell type, observed and
+   generated separately: donor x type summed counts on all 5563 scoring genes (donors of the two groups with
+   >= 1 cell of the type) -> `welch_bh_deg` (log1p CPM, two-sided Welch t-test case vs control, Benjamini-Hochberg over the
+   genes with a defined p-value; needs >= 2 donors per group) -> `top_deg_genes` (FDR < 0.05, descending |logFC|,
+   first 50; shorter when fewer genes pass). `top50_overlap` = |observed list & generated list|; `top50_jaccard`
+   the evaluator's Jaccard. Summary `sum` = total over the scored types (the quantity reported as 81), `max` = the sum of
+   the observed list lengths (the "of 262"), `sum_4type` = T + B + Monocyte + NK. A type is scored when its observed test
+   is defined; an arm whose generated test is undefined for a scored type gets overlap 0. Types: T cell, B cell, Monocyte, NK cell, DC
+   (aggregates: /home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scoring_cache/giroux2022/deg_reference.npz).
+   **Broadcast analogue** (`top50_overlap_broadcast`, as `sle_broadcast.py` `broadcast_input_bulk`; not part of the
+   published benchmark): the same test on the arm's condition bulk_counts (scoring genes, the 6 + 6
+   donors), one whole-PBMC top-50 list, compared with every type's observed list.
+
+## Files
+- `per_donor.csv`: one row per arm x donor (all per-donor quantities above).
+- `summary.json`: per arm the donor means / pooled values, provenance (NPZ sha, receipt check, generation metadata,
+  Poisson nnz of the whole NPZ, annotation contract), and the reference block (PCA, R2 control).
+- `contrast.csv`: per arm x cell type x statistic (`mean_cp10k`, `pseudobulk_log1p_cpm`: logFC r; `welch_top50_deg`:
+  top-50 DEG overlap, T / B / Monocyte / NK / DC; columns of the other kind are empty).
+- `arms/<arm>/labels/` (predict_scanvi.py labels + manifest), `arms/<arm>/generated_cells.parquet` (NPZ row, donor, label,
+  UMI on the scoring genes, P/C subsample flag), `arms/<arm>/deg_top50_lists.csv` (observed, generated and broadcast
+  top-50 gene lists, Ensembl ids, rank order).

paper-releases/pbisc-diffusion/updates/20261003T174003Z-scsemi-readout-sensitivity/diffusion/dataset/server_1/model8/plans/paired_finetune_intradataset_20260927/scores/giroux2022/fewshot_n5/draw2_scsemi5nt/contrast.csv

@@ -0,0 +1,14 @@
+arm,celltype,statistic,n_genes,r_generated,r_broadcast,partial_r_generated_given_broadcast,r_generated_vs_broadcast,n_case_donors_obs,n_control_donors_obs,n_case_donors_gen,n_control_donors_gen,n_cells_obs,n_cells_gen,deg_scored,top50_overlap,top50_jaccard,obs_top_list_size,gen_top_list_size,n_sig_obs,n_sig_gen,top50_overlap_broadcast,top50_jaccard_broadcast,broadcast_top_list_size,n_sig_broadcast
+scsemi5nt,T cell,mean_cp10k,2893,0.5486548838201836,0.45635058627701297,0.40203164614053993,0.5450533859959584,6,6,6,6,67243,9857,,,,,,,,,,,
+scsemi5nt,T cell,pseudobulk_log1p_cpm,2893,0.4743852126136539,0.6892255669260532,0.21848521737778856,0.487782885019727,6,6,6,6,67243,9857,,,,,,,,,,,
+scsemi5nt,B cell,mean_cp10k,3023,0.26976152440798074,0.4100058931750264,0.12945445735002858,0.3931567727243511,6,6,6,6,10555,840,,,,,,,,,,,
+scsemi5nt,B cell,pseudobulk_log1p_cpm,3023,0.15705181140205376,0.6010434257741435,0.02369103743343256,0.23064565693216163,6,6,6,6,10555,840,,,,,,,,,,,
+scsemi5nt,Monocyte,mean_cp10k,3052,0.5235041030230804,0.605855757517743,0.321067266561473,0.498615109517721,6,6,6,6,25096,4429,,,,,,,,,,,
+scsemi5nt,Monocyte,pseudobulk_log1p_cpm,3052,0.2969730035910665,0.7367636063610655,0.12575891241045103,0.2927202411588591,6,6,6,6,25096,4429,,,,,,,,,,,
+scsemi5nt,NK cell,mean_cp10k,2925,0.42607751265995447,0.3796040355648556,0.2934472765238685,0.5052523802334172,6,6,6,6,16282,2167,,,,,,,,,,,
+scsemi5nt,NK cell,pseudobulk_log1p_cpm,2925,0.2859808469877665,0.6122082213455176,0.11488170101295735,0.3269070048496985,6,6,6,6,16282,2167,,,,,,,,,,,
+scsemi5nt,T cell,welch_top50_deg,5563,,,,,6,6,6,6,67243,9857,True,0,0.0,39,1,39,1,6,0.15384615384615385,6,6
+scsemi5nt,B cell,welch_top50_deg,5563,,,,,6,6,6,6,10555,840,True,0,0.0,39,0,39,0,6,0.15384615384615385,6,6
+scsemi5nt,Monocyte,welch_top50_deg,5563,,,,,6,6,6,6,25096,4429,True,0,0.0,9,1,9,1,5,0.5,6,6
+scsemi5nt,NK cell,welch_top50_deg,5563,,,,,6,6,6,6,16282,2167,True,0,0.0,17,1,17,1,5,0.2777777777777778,6,6
+scsemi5nt,DC,welch_top50_deg,5563,,,,,6,6,6,6,1911,70,True,0,0.0,5,0,5,0,4,0.5714285714285714,6,6

paper-releases/pbisc-diffusion/updates/20261003T174003Z-scsemi-readout-sensitivity/diffusion/dataset/server_1/model8/plans/paired_finetune_intradataset_20260927/scores/giroux2022/fewshot_n5/draw2_scsemi5nt/per_donor.csv

@@ -0,0 +1,13 @@
+arm,donor,group,n_generated_cells,n_observed_cells,n_reference_cells,gen_T cell,obs_T cell,gen_B cell,obs_B cell,gen_Monocyte,obs_Monocyte,gen_NK cell,obs_NK cell,gen_DC,obs_DC,gen_Other,obs_Other,comp_l1,comp_l1_obs_loo_null,pb_r,pb_own_rank,pb_top1,precision,coverage,precision_depth_matched,coverage_depth_matched,mono_pct_gen,mono_pct_obs,median_umi_gen,median_umi_obs
+scsemi5nt,7085CA_d0,mild,1500,12585,150,0.19733333333333333,0.21660707191100517,0.021333333333333333,0.024791418355184745,0.6046666666666667,0.5874453714739769,0.14,0.13198251887167262,0.002,0.017719507350019866,0.034666666666666665,0.021454112038140644,0.07690266189908632,0.8990236963821155,0.9398966938463368,1,True,0.9933333333333333,0.6866666666666666,0.98,0.9133333333333333,60.46666666666667,58.74453714739769,447.0,640.0
+scsemi5nt,7085CA_d7,mild,1500,9515,150,0.5106666666666667,0.521492380451918,0.03133333333333333,0.0722017866526537,0.20933333333333334,0.10110352075669994,0.22,0.2725170782974251,0.008666666666666666,0.019863373620599055,0.02,0.012821860220704152,0.23081590471185842,0.3579081617616235,0.9584654064923805,1,True,1.0,0.9,1.0,0.9,20.933333333333334,10.110352075669994,446.0,502.0
+scsemi5nt,7085CA_d14,mild,1500,11887,150,0.5226666666666666,0.472533019264743,0.035333333333333335,0.07133843694792631,0.216,0.16210986792294102,0.19933333333333333,0.24270211155043325,0.006666666666666667,0.018591738874400604,0.02,0.032724825439555814,0.2080475589579652,0.29573495877524286,0.963138696725752,2,False,1.0,0.8066666666666666,1.0,0.8866666666666667,21.6,16.2109867922941,448.0,536.0
+scsemi5nt,2AD75E_d0,mild,1500,7693,150,0.6493333333333333,0.6034056934875861,0.036,0.06408423241908229,0.164,0.14064734173924348,0.12933333333333333,0.15949564539191471,0.0013333333333333333,0.025347718705316522,0.02,0.007019368256856883,0.16452185969929373,0.26792313410215146,0.9318970344966772,4,False,1.0,0.8666666666666667,0.9933333333333333,0.86,16.400000000000002,14.064734173924348,448.0,532.0
+scsemi5nt,2AD75E_d7,mild,1500,10499,150,0.6866666666666666,0.5808172206876845,0.04133333333333333,0.14372797409277074,0.14133333333333334,0.10277169254214687,0.11066666666666666,0.14715687208305553,0.004666666666666667,0.013715591961139155,0.015333333333333332,0.011810648633203162,0.2958675429405976,0.3201835922843184,0.9401929319635772,1,True,1.0,0.8733333333333333,1.0,0.8866666666666667,14.133333333333335,10.277169254214687,447.0,443.0
+scsemi5nt,2AD75E_d14,mild,1500,13692,150,0.6886666666666666,0.5811422728600643,0.043333333333333335,0.12503651767455448,0.14933333333333335,0.14198071866783524,0.09266666666666666,0.12496348232544552,0.006,0.016579024247735902,0.02,0.010297984224364592,0.24915804849547182,0.24443705257466458,0.9372081666715075,1,True,1.0,0.9133333333333333,1.0,0.9466666666666667,14.933333333333335,14.198071866783524,448.0,428.0
+scsemi5nt,82CCF5_d0,severe,1500,10125,150,0.5546666666666666,0.5422222222222223,0.11933333333333333,0.11832098765432099,0.18933333333333333,0.1539753086419753,0.116,0.09303703703703704,0.004,0.016691358024691356,0.016666666666666666,0.07575308641975309,0.1435555555555555,0.23874194415185002,0.9510575879151822,7,False,1.0,0.86,1.0,0.9,18.933333333333334,15.397530864197531,447.0,390.0
+scsemi5nt,82CCF5_d7,severe,1500,10860,150,0.6366666666666667,0.619060773480663,0.056,0.09972375690607735,0.14533333333333334,0.14152854511970533,0.094,0.09410681399631676,0.002,0.013627992633517496,0.066,0.03195211786372008,0.11091712707182325,0.26485931945363844,0.9496336124059367,2,False,1.0,0.92,1.0,0.94,14.533333333333335,14.152854511970533,447.0,497.0
+scsemi5nt,82CCF5_d14,severe,1500,11056,150,0.676,0.6330499276410999,0.044,0.10121201157742403,0.13,0.11369392185238784,0.096,0.06828871201157742,0.0026666666666666666,0.014562228654124458,0.051333333333333335,0.0691931982633864,0.17393487698986976,0.36985318969456304,0.9507546301905615,1,True,1.0,0.8533333333333334,1.0,0.88,13.0,11.369392185238784,447.0,394.0
+scsemi5nt,0BF51C_d0,severe,1500,6510,150,0.25,0.4012288786482335,0.03333333333333333,0.01674347158218126,0.532,0.4609831029185868,0.14,0.07726574500768049,0.0006666666666666666,0.006144393241167435,0.044,0.03763440860215054,0.3134132104454685,0.5709498030244532,0.9298562932480927,1,True,0.9933333333333333,0.7133333333333334,0.88,0.72,53.2,46.09831029185868,448.0,628.0
+scsemi5nt,0BF51C_d7,severe,1500,10105,150,0.5606666666666666,0.612370113805047,0.05466666666666667,0.05700148441365661,0.258,0.1607125185551707,0.056,0.05294408708560119,0.0013333333333333333,0.0073231073725878275,0.06933333333333333,0.10964868876793667,0.20068678871845624,0.3232914310644849,0.9219086006135095,6,False,1.0,0.8466666666666667,1.0,0.8866666666666667,25.8,16.071251855517072,448.0,509.0
+scsemi5nt,0BF51C_d14,severe,1500,11419,150,0.638,0.6330676941938874,0.044,0.07960416849111131,0.21333333333333335,0.151589456169542,0.050666666666666665,0.07907872843506436,0.006666666666666667,0.010508801120938787,0.04733333333333333,0.04615115158945617,0.13571672942756222,0.23769151748520465,0.9411979085590674,3,False,1.0,0.7866666666666666,1.0,0.9333333333333333,21.333333333333336,15.1589456169542,448.0,541.0

paper-releases/pbisc-diffusion/updates/20261003T174003Z-scsemi-readout-sensitivity/diffusion/dataset/server_1/model8/plans/paired_finetune_intradataset_20260927/scores/giroux2022/fewshot_n5/draw2_scsemi5nt/summary.json

@@ -0,0 +1,483 @@
+{
+ "schema": "paired_ft_scores.v1",
+ "dataset": "giroux2022",
+ "written_utc": "2026-10-03T15:03:02.132776+00:00",
+ "command": [
+  "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/tools/score_ft.py",
+  "--dataset",
+  "giroux2022",
+  "--out",
+  "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores/giroux2022/fewshot_n5/draw2_scsemi5nt",
+  "--threads",
+  "6",
+  "--donors",
+  "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data/giroux2022/fewshot_n5/draw2/test_samples.txt",
+  "--case-groups",
+  "severe",
+  "--control-groups",
+  "mild",
+  "--arm",
+  "scsemi5nt=/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/generated/giroux2022/fewshot_n5/draw2/scsemi5nt__raw.npz"
+ ],
+ "arms": [
+  "scsemi5nt"
+ ],
+ "donors": [
+  "7085CA_d0",
+  "7085CA_d7",
+  "7085CA_d14",
+  "2AD75E_d0",
+  "2AD75E_d7",
+  "2AD75E_d14",
+  "82CCF5_d0",
+  "82CCF5_d7",
+  "82CCF5_d14",
+  "0BF51C_d0",
+  "0BF51C_d7",
+  "0BF51C_d14"
+ ],
+ "n_donors": 12,
+ "reference": {
+  "cache": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scoring_cache/giroux2022",
+  "reference_sha256": "dd0152f7bde41da80e1c71dabf54ee561982f36dc99a336b65fca419880a1593",
+  "features_sha256": "fb41b91617e3ff786d5153c4dc7869d66191ca5947ff28161d24215a24f819e5",
+  "n_r1_cells": 1800,
+  "pca": {
+   "n_components": 50,
+   "explained_variance_ratio_sum": 0.2604835629463196,
+   "n_fit_cells": 1800
+  },
+  "r2_control": {
+   "precision": 0.9988888888888889,
+   "coverage": 1.0,
+   "density": 4.929388888888889,
+   "n_query": 1800,
+   "depth_matched": {
+    "precision": 0.9994444444444445,
+    "coverage": 1.0,
+    "density": 4.851333333333334
+   }
+  },
+  "observed_l1_loo_null_mean": 0.36588315006285926
+ },
+ "contrast_definition": {
+  "enabled": true,
+  "rule": "explicit --case-groups / --control-groups",
+  "group_source": "features groups",
+  "case_groups": [
+   "severe"
+  ],
+  "control_groups": [
+   "mild"
+  ],
+  "n_case_donors": 6,
+  "n_control_donors": 6,
+  "n_donors_in_neither": 0,
+  "case": [
+   false,
+   false,
+   false,
+   false,
+   false,
+   false,
+   true,
+   true,
+   true,
+   true,
+   true,
+   true
+  ],
+  "ctrl": [
+   true,
+   true,
+   true,
+   true,
+   true,
+   true,
+   false,
+   false,
+   false,
+   false,
+   false,
+   false
+  ],
+  "groups": [
+   "mild",
+   "mild",
+   "mild",
+   "mild",
+   "mild",
+   "mild",
+   "severe",
+   "severe",
+   "severe",
+   "severe",
+   "severe",
+   "severe"
+  ]
+ },
+ "deg_definition": {
+  "enabled": true,
+  "score": "top-50 DEG overlap per cell type (benchmark total_top50_overlap, Welch backend)",
+  "source": "/home/sj_server_1/PBISC/immun-pbmc-fm/bulk2sc-paper/scripts/evaluate_generated_vs_truth.py",
+  "functions": [
+   "welch_bh_deg",
+   "top_deg_genes",
+   "jaccard"
+  ],
+  "reproduces": "utility_studies_final6k_20260925/sle top-50 DEG 81 (1,006 genes) / 84 (2,279 genes) = sum over the six gated types of top50_overlap_count",
+  "top_n": 50,
+  "fdr": 0.05,
+  "n_genes": 5563,
+  "types": [
+   "T cell",
+   "B cell",
+   "Monocyte",
+   "NK cell",
+   "DC"
+  ],
+  "aggregates_source": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scoring_cache/giroux2022/deg_reference.npz",
+  "aggregates_sha256": "412c44bba7481b0b851c3b12e695f1b7ff662163e1f7849d3a64359dfce6d50e",
+  "aggregates_note": null,
+  "types_not_scored_observed": [],
+  "observed": {
+   "T cell": {
+    "n_significant": 39,
+    "n_tested": 5281,
+    "n_case_donors": 6,
+    "n_control_donors": 6,
+    "n_cells": 67243,
+    "top_list_size": 39
+   },
+   "B cell": {
+    "n_significant": 39,
+    "n_tested": 5004,
+    "n_case_donors": 6,
+    "n_control_donors": 6,
+    "n_cells": 10555,
+    "top_list_size": 39
+   },
+   "Monocyte": {
+    "n_significant": 9,
+    "n_tested": 5206,
+    "n_case_donors": 6,
+    "n_control_donors": 6,
+    "n_cells": 25096,
+    "top_list_size": 9
+   },
+   "NK cell": {
+    "n_significant": 17,
+    "n_tested": 5018,
+    "n_case_donors": 6,
+    "n_control_donors": 6,
+    "n_cells": 16282,
+    "top_list_size": 17
+   },
+   "DC": {
+    "n_significant": 5,
+    "n_tested": 4736,
+    "n_case_donors": 6,
+    "n_control_donors": 6,
+    "n_cells": 1911,
+    "top_list_size": 5
+   }
+  },
+  "broadcast": "welch_bh_deg + top_deg_genes on the arm's condition bulk_counts (scoring genes, case|control donors), one list compared with every type's observed list (sle_broadcast.py broadcast_input_bulk); not part of the published benchmark"
+ },
+ "headline": {
+  "scsemi5nt": {
+   "comp_l1": 0.19196148874275074,
+   "comp_l1_obs_loo_null": 0.36588315006285926,
+   "precision": 0.9988888888888889,
+   "coverage": 0.8355555555555556,
+   "precision_dm": 0.9877777777777778,
+   "coverage_dm": 0.8877777777777778,
+   "pb_r_median": 0.9406954202613222,
+   "pb_top1": 0.5,
+   "mono_pct_gen": 24.605555555555554,
+   "mono_pct_obs": 20.154511386335095,
+   "median_umi_gen": 447.0,
+   "median_umi_obs": 495.0,
+   "contrast_mean_cp10k_r_gen": 0.4419995059777998,
+   "contrast_pseudobulk_log1p_cpm_r_gen": 0.30359771864863516,
+   "contrast_mean_cp10k_r_broadcast": 0.46295406813365947,
+   "contrast_pseudobulk_log1p_cpm_r_broadcast": 0.659810205101695,
+   "deg_top50_T": 0,
+   "deg_top50_B": 0,
+   "deg_top50_Mono": 0,
+   "deg_top50_NK": 0,
+   "deg_top50_DC": 0,
+   "deg_top50_sum": 0,
+   "deg_top50_max": 109,
+   "deg_top50_sum_4type": 0,
+   "deg_top50_sum_broadcast": 26,
+   "seconds": 6.9
+  }
+ },
+ "per_arm": {
+  "scsemi5nt": {
+   "arm": "scsemi5nt",
+   "npz": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/generated/giroux2022/fewshot_n5/draw2/scsemi5nt__raw.npz",
+   "npz_sha256": "d54c7c87be660ada0688e0f1b25ebdc31e97107fdc6a36d43c13630de83d210a",
+   "receipt_verified": true,
+   "generation": {
+    "checkpoint": null,
+    "checkpoint_sha256": null,
+    "input_condition_path": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data/giroux2022/fewshot_n5/draw2/test_raw.npz",
+    "input_condition_sha256": "fedb8e8039576499367bfec5479e15b4bac720126d8c5d7412d1d12f85531c28",
+    "seed": 20260927,
+    "steps": null,
+    "support_readout": "no output threshold: the upstream model's ZINB mean on every modeled gene (sample(bound=0), no xsemi > 10 step); read-out sensitivity variant, not the official output",
+    "weights": null
+   },
+   "condition_for_broadcast": {
+    "path": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data/giroux2022/fewshot_n5/draw2/test_raw.npz",
+    "sha256": "fedb8e8039576499367bfec5479e15b4bac720126d8c5d7412d1d12f85531c28"
+   },
+   "poisson": {
+    "seed": 20260916,
+    "rule": "whole NPZ, rows in NPZ order",
+    "nnz_whole_npz": 3830523,
+    "median_library_whole_npz_panel": 447.0,
+    "n_rows_whole_npz": 18000
+   },
+   "annotation": {
+    "model_pt_sha256": "bf3655899e5abee435340fc008656c4a17342da0f30b70cfd43de8e9223e1930",
+    "zero_padded_model_genes": 0,
+    "label_counts": {
+     "T cell": 9857,
+     "Monocyte": 4429,
+     "NK cell": 2167,
+     "B cell": 840,
+     "Other": 637,
+     "DC": 70
+    },
+    "mean_l2_max_prob": 0.8149558672250973,
+    "predict_seconds": 3.6
+   },
+   "n_donors": 12,
+   "n_generated_cells": 18000,
+   "composition": {
+    "l1_mean": 0.19196148874275074,
+    "l1_median": 0.187310832854163,
+    "l1_obs_loo_null_mean": 0.36588315006285926,
+    "l1_generated_loo_mean_vs_observed_mean": 0.4010839403973465,
+    "composition_top1": 0.3333333333333333,
+    "mean_fraction_generated": {
+     "T cell": 0.547611,
+     "B cell": 0.046667,
+     "Monocyte": 0.246056,
+     "NK cell": 0.120389,
+     "DC": 0.003889,
+     "Other": 0.035389
+    },
+    "mean_fraction_observed": {
+     "T cell": 0.53475,
+     "B cell": 0.081149,
+     "Monocyte": 0.201545,
+     "NK cell": 0.128628,
+     "DC": 0.015056,
+     "Other": 0.038872
+    }
+   },
+   "pseudobulk": {
+    "n_genes": 5563,
+    "r_median": 0.9406954202613222,
+    "r_mean": 0.9429339635940485,
+    "top1": 0.5,
+    "n_top1": 6,
+    "own_rank_median": 1.5,
+    "chance_top1": 0.08333333333333333
+   },
+   "precision_coverage": {
+    "precision": 0.9988888888888889,
+    "coverage": 0.8355555555555556,
+    "density": 3.7124444444444444,
+    "n_query": 1800,
+    "n_reference": 1800
+   },
+   "precision_coverage_depth_matched": {
+    "precision": 0.9877777777777778,
+    "coverage": 0.8877777777777778,
+    "density": 2.6559444444444447,
+    "detected_median_before": {
+     "R1": 257.0,
+     "Q": 215.0
+    },
+    "detected_median_after": {
+     "R1": 212.0,
+     "Q": 212.0
+    }
+   },
+   "monocyte_pct": {
+    "generated_mean_over_donors": 24.605555555555554,
+    "observed_mean_over_donors": 20.154511386335095,
+    "generated_pooled_cells": 24.605555555555554
+   },
+   "median_umi_per_cell_scoring_genes": {
+    "generated_pooled": 447.0,
+    "observed_pooled": 495.0,
+    "generated_mean_of_donor_medians": 447.4166564941406,
+    "observed_mean_of_donor_medians": 503.3333333333333
+   },
+   "contrast": {
+    "mean_cp10k": {
+     "per_type": {
+      "T cell": {
+       "r_generated": 0.5486548838201836,
+       "r_broadcast": 0.45635058627701297,
+       "partial_r_generated_given_broadcast": 0.40203164614053993,
+       "r_generated_vs_broadcast": 0.5450533859959584,
+       "n_genes": 2893
+      },
+      "B cell": {
+       "r_generated": 0.26976152440798074,
+       "r_broadcast": 0.4100058931750264,
+       "partial_r_generated_given_broadcast": 0.12945445735002858,
+       "r_generated_vs_broadcast": 0.3931567727243511,
+       "n_genes": 3023
+      },
+      "Monocyte": {
+       "r_generated": 0.5235041030230804,
+       "r_broadcast": 0.605855757517743,
+       "partial_r_generated_given_broadcast": 0.321067266561473,
+       "r_generated_vs_broadcast": 0.498615109517721,
+       "n_genes": 3052
+      },
+      "NK cell": {
+       "r_generated": 0.42607751265995447,
+       "r_broadcast": 0.3796040355648556,
+       "partial_r_generated_given_broadcast": 0.2934472765238685,
+       "r_generated_vs_broadcast": 0.5052523802334172,
+       "n_genes": 2925
+      }
+     },
+     "mean_r_generated": 0.4419995059777998,
+     "mean_r_broadcast": 0.46295406813365947,
+     "mean_partial_r": 0.2865001616439775
+    },
+    "pseudobulk_log1p_cpm": {
+     "per_type": {
+      "T cell": {
+       "r_generated": 0.4743852126136539,
+       "r_broadcast": 0.6892255669260532,
+       "partial_r_generated_given_broadcast": 0.21848521737778856,
+       "r_generated_vs_broadcast": 0.487782885019727,
+       "n_genes": 2893
+      },
+      "B cell": {
+       "r_generated": 0.15705181140205376,
+       "r_broadcast": 0.6010434257741435,
+       "partial_r_generated_given_broadcast": 0.02369103743343256,
+       "r_generated_vs_broadcast": 0.23064565693216163,
+       "n_genes": 3023
+      },
+      "Monocyte": {
+       "r_generated": 0.2969730035910665,
+       "r_broadcast": 0.7367636063610655,
+       "partial_r_generated_given_broadcast": 0.12575891241045103,
+       "r_generated_vs_broadcast": 0.2927202411588591,
+       "n_genes": 3052
+      },
+      "NK cell": {
+       "r_generated": 0.2859808469877665,
+       "r_broadcast": 0.6122082213455176,
+       "partial_r_generated_given_broadcast": 0.11488170101295735,
+       "r_generated_vs_broadcast": 0.3269070048496985,
+       "n_genes": 2925
+      }
+     },
+     "mean_r_generated": 0.30359771864863516,
+     "mean_r_broadcast": 0.659810205101695,
+     "mean_partial_r": 0.12070421705865739
+    }
+   },
+   "deg_top50": {
+    "types_scored": [
+     "T cell",
+     "B cell",
+     "Monocyte",
+     "NK cell",
+     "DC"
+    ],
+    "types_not_scored": {},
+    "per_type": {
+     "T cell": {
+      "deg_scored": true,
+      "top50_overlap": 0,
+      "top50_jaccard": 0.0,
+      "obs_top_list_size": 39,
+      "gen_top_list_size": 1,
+      "top50_overlap_broadcast": 6,
+      "top50_jaccard_broadcast": 0.15384615384615385,
+      "n_cells_obs": 67243,
+      "n_cells_gen": 9857
+     },
+     "B cell": {
+      "deg_scored": true,
+      "top50_overlap": 0,
+      "top50_jaccard": 0.0,
+      "obs_top_list_size": 39,
+      "gen_top_list_size": 0,
+      "top50_overlap_broadcast": 6,
+      "top50_jaccard_broadcast": 0.15384615384615385,
+      "n_cells_obs": 10555,
+      "n_cells_gen": 840
+     },
+     "Monocyte": {
+      "deg_scored": true,
+      "top50_overlap": 0,
+      "top50_jaccard": 0.0,
+      "obs_top_list_size": 9,
+      "gen_top_list_size": 1,
+      "top50_overlap_broadcast": 5,
+      "top50_jaccard_broadcast": 0.5,
+      "n_cells_obs": 25096,
+      "n_cells_gen": 4429
+     },
+     "NK cell": {
+      "deg_scored": true,
+      "top50_overlap": 0,
+      "top50_jaccard": 0.0,
+      "obs_top_list_size": 17,
+      "gen_top_list_size": 1,
+      "top50_overlap_broadcast": 5,
+      "top50_jaccard_broadcast": 0.2777777777777778,
+      "n_cells_obs": 16282,
+      "n_cells_gen": 2167
+     },
+     "DC": {
+      "deg_scored": true,
+      "top50_overlap": 0,
+      "top50_jaccard": 0.0,
+      "obs_top_list_size": 5,
+      "gen_top_list_size": 0,
+      "top50_overlap_broadcast": 4,
+      "top50_jaccard_broadcast": 0.5714285714285714,
+      "n_cells_obs": 1911,
+      "n_cells_gen": 70
+     }
+    },
+    "sum": 0,
+    "sum_4type": 0,
+    "max": 109,
+    "mean": 0.0,
+    "sum_broadcast": 26,
+    "sum_broadcast_4type": 22,
+    "broadcast_n_significant": 6,
+    "seconds": 0.04
+   },
+   "seconds": 6.9
+  }
+ },
+ "code_sha256": {
+  "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/tools/score_ft.py": "ec5f40b4cea6d5062d0138f3b934abb6465e34cbfeac869c8a0906924c033aec",
+  "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/tools/ft_common.py": "4a6366b0f22208cd58714fc0261b8840c643b9d4569704d560cfd93534cbc747",
+  "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/utility_studies_final6k_20260925/umap/cell_immicro_5arm/tools/stage_b_embed.py": "d056cfdf38fb0e5f478756f370174e993b75d0c973c48e02e21749175a4ca2da",
+  "/home/sj_server_1/PBISC/immun-pbmc-fm/bulk2sc-paper/scripts/evaluate_generated_vs_truth.py": "48b12d794782b840838853e93e0fc55fab00532c4b3698f4588e926b985d2881",
+  "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/utility_studies_final6k_20260925/immicro/tools/immicro_common.py": "c5c30e917148cc5e664e91ea30c11bbdff9b6d1f53a3e0a13e16b1917eff37e9",
+  "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/aifi_scanvi_rebuild_20260916/predict_scanvi.py": "482ee5dcf4d72b54bc36cbb6dcff62e5030bc7afc46cb5fb01eca73a82ee45a5",
+  "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/aifi_scanvi_rebuild_20260916/poisson_readout.py": "8bfa080942fbcdc89ae23b276524a1a83266fb8cbc56b9af279e50f82e6cde38"
+ },
+ "seconds": 7.7
+}