Archive 4 checkpoints (1-4)

paper-releases/pbisc-diffusion/updates/20260930T065633Z-summary-deck-scsemi-final/diffusion/dataset/server_1/model8/plans/paired_finetune_intradataset_20260927/scores/giroux2022/fewshot_n5/draw3_scsemi5/README.md

@@ -0,0 +1,73 @@
+# score_ft.py output: giroux2022, 11 scored donors
+
+Arms: scsemi5.  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 11 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) (1650 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 0.9988, 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 = 5); 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 + 5
+   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/20260930T065633Z-summary-deck-scsemi-final/diffusion/dataset/server_1/model8/plans/paired_finetune_intradataset_20260927/scores/giroux2022/fewshot_n5/draw3_scsemi5/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
+scsemi5,T cell,mean_cp10k,2907,0.4955015640545308,0.5426693698646771,0.3483993335853783,0.4249364087840753,6,5,6,5,69840,9625,,,,,,,,,,,
+scsemi5,T cell,pseudobulk_log1p_cpm,2907,0.39932987372898515,0.7425337537971076,0.15152247412519773,0.41333363600083306,6,5,6,5,69840,9625,,,,,,,,,,,
+scsemi5,B cell,mean_cp10k,3041,0.23843165332677707,0.46011992006345803,0.12854750170213747,0.28007397801020556,6,5,6,5,11400,650,,,,,,,,,,,
+scsemi5,B cell,pseudobulk_log1p_cpm,3041,0.16560878455733485,0.6652140288020838,0.020652986462832576,0.22637600606689282,6,5,6,5,11400,650,,,,,,,,,,,
+scsemi5,Monocyte,mean_cp10k,3060,0.3557163348952947,0.5915210794160278,0.1550721771974655,0.4084159697412405,6,5,6,5,18916,3890,,,,,,,,,,,
+scsemi5,Monocyte,pseudobulk_log1p_cpm,3060,0.2402243450115651,0.7553162442814422,0.03987684856844198,0.28487871581989443,6,5,6,5,18916,3890,,,,,,,,,,,
+scsemi5,NK cell,mean_cp10k,2949,0.37319733443777137,0.45003670363580767,0.24635767530306976,0.37635295569640487,6,5,6,5,12021,1769,,,,,,,,,,,
+scsemi5,NK cell,pseudobulk_log1p_cpm,2949,0.232791516548745,0.6686724823885001,0.045479831025545675,0.29989446409205933,6,5,6,5,12021,1769,,,,,,,,,,,
+scsemi5,T cell,welch_top50_deg,5563,,,,,6,5,6,5,69840,9625,True,2,0.05,39,3,39,3,5,0.12195121951219512,7,7
+scsemi5,B cell,welch_top50_deg,5563,,,,,6,5,6,5,11400,650,True,0,0.0,17,1,17,1,5,0.2631578947368421,7,7
+scsemi5,Monocyte,welch_top50_deg,5563,,,,,6,5,6,5,18916,3890,True,0,0.0,6,0,6,0,5,0.625,7,7
+scsemi5,NK cell,welch_top50_deg,5563,,,,,6,5,6,5,12021,1769,True,1,0.1111111111111111,9,1,9,1,5,0.45454545454545453,7,7
+scsemi5,DC,welch_top50_deg,5563,,,,,6,5,6,5,1706,68,True,0,0.0,4,0,4,0,4,0.5714285714285714,7,7

paper-releases/pbisc-diffusion/updates/20260930T065633Z-summary-deck-scsemi-final/diffusion/dataset/server_1/model8/plans/paired_finetune_intradataset_20260927/scores/giroux2022/fewshot_n5/draw3_scsemi5/per_donor.csv

@@ -0,0 +1,12 @@
+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
+scsemi5,0B943B_d0,mild,1500,11344,150,0.5026666666666667,0.6220028208744711,0.028666666666666667,0.08312764456981664,0.2773333333333333,0.12588152327221439,0.174,0.11865303244005641,0.008666666666666666,0.023977433004231313,0.008666666666666666,0.026357545839210156,0.413597555242125,0.14544387566831765,0.5925792096451624,9,False,0.9933333333333333,0.43333333333333335,0.4866666666666667,0.08,27.73333333333333,12.588152327221438,463.0,539.0
+scsemi5,0B943B_d14,mild,1500,15650,150,0.424,0.5652396166134185,0.036,0.11175718849840256,0.3293333333333333,0.17086261980830672,0.156,0.09789137380191694,0.004666666666666667,0.009968051118210863,0.05,0.04428115015974441,0.44459637912673056,0.04452446481654314,0.5922307390103829,6,False,0.9866666666666667,0.6933333333333334,0.6066666666666667,0.2733333333333333,32.93333333333333,17.08626198083067,462.0,461.0
+scsemi5,2AD75E_d0,mild,1500,7693,150,0.6693333333333333,0.6034056934875861,0.035333333333333335,0.06408423241908229,0.164,0.14064734173924348,0.11533333333333333,0.15949564539191471,0.0013333333333333333,0.025347718705316522,0.014666666666666666,0.007019368256856883,0.19385519303262708,0.22342037955369395,0.5907116929908761,3,False,0.9933333333333333,0.58,0.44666666666666666,0.2,16.400000000000002,14.064734173924348,462.0,532.0
+scsemi5,2AD75E_d7,mild,1500,10499,150,0.696,0.5808172206876845,0.042666666666666665,0.14372797409277074,0.14066666666666666,0.10277169254214687,0.104,0.14715687208305553,0.004666666666666667,0.013715591961139155,0.012,0.011810648633203162,0.3065342096072642,0.2508849036101115,0.6089825751862754,1,True,1.0,0.72,0.5133333333333333,0.2733333333333333,14.066666666666666,10.277169254214687,461.0,443.0
+scsemi5,2AD75E_d14,mild,1500,13692,150,0.692,0.5811422728600643,0.043333333333333335,0.12503651767455448,0.15133333333333332,0.14198071866783524,0.08866666666666667,0.12496348232544552,0.006,0.016579024247735902,0.018666666666666668,0.010297984224364592,0.25715804849547175,0.17549218017641194,0.5930078380086253,2,False,1.0,0.78,0.5066666666666667,0.14666666666666667,15.133333333333333,14.198071866783524,462.0,428.0
+scsemi5,82CCF5_d0,severe,1500,10125,150,0.4886666666666667,0.5422222222222223,0.02266666666666667,0.11832098765432099,0.30133333333333334,0.1539753086419753,0.14266666666666666,0.09303703703703704,0.006,0.016691358024691356,0.03866666666666667,0.07575308641975309,0.3939753086419753,0.1495689270510743,0.6209210359420592,6,False,1.0,0.5866666666666667,0.5266666666666666,0.24,30.133333333333333,15.397530864197531,463.0,390.0
+scsemi5,82CCF5_d7,severe,1500,10860,150,0.6633333333333333,0.619060773480663,0.056,0.09972375690607735,0.14466666666666667,0.14152854511970533,0.086,0.09410681399631676,0.0013333333333333333,0.013627992633517496,0.048666666666666664,0.03195211786372008,0.12825046040515653,0.11747981254548856,0.6849841617396683,1,True,0.9933333333333333,0.7133333333333334,0.26,0.3,14.466666666666667,14.152854511970533,463.0,497.0
+scsemi5,82CCF5_d14,severe,1500,11056,150,0.712,0.6330499276410999,0.042,0.10121201157742403,0.13,0.11369392185238784,0.084,0.06828871201157742,0.002,0.014562228654124458,0.03,0.0691931982633864,0.2219348769898697,0.2142542173330148,0.67053556522553,2,False,0.9933333333333333,0.7,0.17333333333333334,0.23333333333333334,13.0,11.369392185238784,462.0,394.0
+scsemi5,0BF51C_d0,severe,1500,6510,150,0.286,0.4012288786482335,0.034,0.01674347158218126,0.506,0.4609831029185868,0.13666666666666666,0.07726574500768049,0.0013333333333333333,0.006144393241167435,0.036,0.03763440860215054,0.24334869431643624,0.6552333302154613,0.6478705401656658,1,True,0.9333333333333333,0.4666666666666667,0.4266666666666667,0.38666666666666666,50.6,46.09831029185868,461.0,628.0
+scsemi5,0BF51C_d7,severe,1500,10105,150,0.606,0.612370113805047,0.050666666666666665,0.05700148441365661,0.24533333333333332,0.1607125185551707,0.04733333333333333,0.05294408708560119,0.002,0.0073231073725878275,0.048666666666666664,0.10964868876793667,0.16924162955632527,0.2367258124634929,0.5705712193060504,6,False,1.0,0.6466666666666666,0.49333333333333335,0.17333333333333334,24.53333333333333,16.071251855517072,462.0,509.0
+scsemi5,0BF51C_d14,severe,1500,11419,150,0.6766666666666666,0.6330676941938874,0.042,0.07960416849111131,0.20333333333333334,0.151589456169542,0.04466666666666667,0.07907872843506436,0.007333333333333333,0.010508801120938787,0.026,0.04615115158945617,0.1906856992731412,0.13287977411295523,0.6255747592802042,4,False,0.9866666666666667,0.6,0.42,0.24666666666666667,20.333333333333332,15.1589456169542,462.0,541.0

paper-releases/pbisc-diffusion/updates/20260930T065633Z-summary-deck-scsemi-final/diffusion/dataset/server_1/model8/plans/paired_finetune_intradataset_20260927/scores/giroux2022/fewshot_n5/draw3_scsemi5/summary.json

@@ -0,0 +1,479 @@
+{
+ "schema": "paired_ft_scores.v1",
+ "dataset": "giroux2022",
+ "written_utc": "2026-09-29T21:06:25.532952+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/draw3_scsemi5",
+  "--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/draw3/test_samples.txt",
+  "--case-groups",
+  "severe",
+  "--control-groups",
+  "mild",
+  "--arm",
+  "scsemi5=/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/generated/giroux2022/fewshot_n5/draw3/scsemi5__raw.npz"
+ ],
+ "arms": [
+  "scsemi5"
+ ],
+ "donors": [
+  "0B943B_d0",
+  "0B943B_d14",
+  "2AD75E_d0",
+  "2AD75E_d7",
+  "2AD75E_d14",
+  "82CCF5_d0",
+  "82CCF5_d7",
+  "82CCF5_d14",
+  "0BF51C_d0",
+  "0BF51C_d7",
+  "0BF51C_d14"
+ ],
+ "n_donors": 11,
+ "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": 1650,
+  "pca": {
+   "n_components": 50,
+   "explained_variance_ratio_sum": 0.2550526559352875,
+   "n_fit_cells": 1650
+  },
+  "r2_control": {
+   "precision": 0.9987878787878788,
+   "coverage": 1.0,
+   "density": 5.4827272727272724,
+   "n_query": 1650,
+   "depth_matched": {
+    "precision": 0.9993939393939394,
+    "coverage": 1.0,
+    "density": 5.389454545454546
+   }
+  },
+  "observed_l1_loo_null_mean": 0.213264334322415
+ },
+ "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": 5,
+  "n_donors_in_neither": 0,
+  "case": [
+   false,
+   false,
+   false,
+   false,
+   false,
+   true,
+   true,
+   true,
+   true,
+   true,
+   true
+  ],
+  "ctrl": [
+   true,
+   true,
+   true,
+   true,
+   true,
+   false,
+   false,
+   false,
+   false,
+   false,
+   false
+  ],
+  "groups": [
+   "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": 5283,
+    "n_case_donors": 6,
+    "n_control_donors": 5,
+    "n_cells": 69840,
+    "top_list_size": 39
+   },
+   "B cell": {
+    "n_significant": 17,
+    "n_tested": 5034,
+    "n_case_donors": 6,
+    "n_control_donors": 5,
+    "n_cells": 11400,
+    "top_list_size": 17
+   },
+   "Monocyte": {
+    "n_significant": 6,
+    "n_tested": 5160,
+    "n_case_donors": 6,
+    "n_control_donors": 5,
+    "n_cells": 18916,
+    "top_list_size": 6
+   },
+   "NK cell": {
+    "n_significant": 9,
+    "n_tested": 4956,
+    "n_case_donors": 6,
+    "n_control_donors": 5,
+    "n_cells": 12021,
+    "top_list_size": 9
+   },
+   "DC": {
+    "n_significant": 4,
+    "n_tested": 4690,
+    "n_case_donors": 6,
+    "n_control_donors": 5,
+    "n_cells": 1706,
+    "top_list_size": 4
+   }
+  },
+  "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": {
+  "scsemi5": {
+   "comp_l1": 0.26937982315337483,
+   "comp_l1_obs_loo_null": 0.213264334322415,
+   "precision": 0.9890909090909091,
+   "coverage": 0.6290909090909091,
+   "precision_dm": 0.44181818181818183,
+   "coverage_dm": 0.23212121212121212,
+   "pb_r_median": 0.6089825751862754,
+   "pb_top1": 0.2727272727272727,
+   "mono_pct_gen": 23.575757575757574,
+   "mono_pct_obs": 16.951152266246496,
+   "median_umi_gen": 462.0,
+   "median_umi_obs": 480.0,
+   "contrast_mean_cp10k_r_gen": 0.36571172167859345,
+   "contrast_pseudobulk_log1p_cpm_r_gen": 0.25948862996165756,
+   "contrast_mean_cp10k_r_broadcast": 0.5110867682449927,
+   "contrast_pseudobulk_log1p_cpm_r_broadcast": 0.7079341273172834,
+   "deg_top50_T": 2,
+   "deg_top50_B": 0,
+   "deg_top50_Mono": 0,
+   "deg_top50_NK": 1,
+   "deg_top50_DC": 0,
+   "deg_top50_sum": 3,
+   "deg_top50_max": 75,
+   "deg_top50_sum_4type": 3,
+   "deg_top50_sum_broadcast": 24,
+   "seconds": 4.6
+  }
+ },
+ "per_arm": {
+  "scsemi5": {
+   "arm": "scsemi5",
+   "npz": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/generated/giroux2022/fewshot_n5/draw3/scsemi5__raw.npz",
+   "npz_sha256": "0d75cfbab0da689ad49e5cf70e4d82593be189f2d71607a861712ea8d83791d4",
+   "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/draw3/test_raw.npz",
+    "input_condition_sha256": "4efbb07e4fe684710ed219d75c9a15685f617c761e2401541613e3e424146823",
+    "seed": 20260927,
+    "steps": null,
+    "support_readout": "upstream (values <= 10 normalized units zeroed by fast_semi)",
+    "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/draw3/test_raw.npz",
+    "sha256": "4efbb07e4fe684710ed219d75c9a15685f617c761e2401541613e3e424146823"
+   },
+   "poisson": {
+    "seed": 20260916,
+    "rule": "whole NPZ, rows in NPZ order",
+    "nnz_whole_npz": 1821079,
+    "median_library_whole_npz_panel": 462.0,
+    "n_rows_whole_npz": 16500
+   },
+   "annotation": {
+    "model_pt_sha256": "bf3655899e5abee435340fc008656c4a17342da0f30b70cfd43de8e9223e1930",
+    "zero_padded_model_genes": 0,
+    "label_counts": {
+     "T cell": 9625,
+     "Monocyte": 3890,
+     "NK cell": 1769,
+     "B cell": 650,
+     "Other": 498,
+     "DC": 68
+    },
+    "mean_l2_max_prob": 0.8345846770288367,
+    "predict_seconds": 3.0
+   },
+   "n_donors": 11,
+   "n_generated_cells": 16500,
+   "composition": {
+    "l1_mean": 0.26937982315337483,
+    "l1_median": 0.24334869431643624,
+    "l1_obs_loo_null_mean": 0.213264334322415,
+    "l1_generated_loo_mean_vs_observed_mean": 0.2913910325861255,
+    "composition_top1": 0.2727272727272727,
+    "mean_fraction_generated": {
+     "T cell": 0.583333,
+     "B cell": 0.039394,
+     "Monocyte": 0.235758,
+     "NK cell": 0.107212,
+     "DC": 0.004121,
+     "Other": 0.030182
+    },
+    "mean_fraction_observed": {
+     "T cell": 0.581237,
+     "B cell": 0.09094,
+     "Monocyte": 0.169512,
+     "NK cell": 0.101171,
+     "DC": 0.014404,
+     "Other": 0.042736
+    }
+   },
+   "pseudobulk": {
+    "n_genes": 5563,
+    "r_median": 0.6089825751862754,
+    "r_mean": 0.6179972124091363,
+    "top1": 0.2727272727272727,
+    "n_top1": 3,
+    "own_rank_median": 3.0,
+    "chance_top1": 0.09090909090909091
+   },
+   "precision_coverage": {
+    "precision": 0.9890909090909091,
+    "coverage": 0.6290909090909091,
+    "density": 1.227030303030303,
+    "n_query": 1650,
+    "n_reference": 1650
+   },
+   "precision_coverage_depth_matched": {
+    "precision": 0.44181818181818183,
+    "coverage": 0.23212121212121212,
+    "density": 0.18884848484848485,
+    "detected_median_before": {
+     "R1": 254.0,
+     "Q": 107.0
+    },
+    "detected_median_after": {
+     "R1": 107.0,
+     "Q": 107.0
+    }
+   },
+   "monocyte_pct": {
+    "generated_mean_over_donors": 23.575757575757574,
+    "observed_mean_over_donors": 16.951152266246496,
+    "generated_pooled_cells": 23.575757575757574
+   },
+   "median_umi_per_cell_scoring_genes": {
+    "generated_pooled": 462.0,
+    "observed_pooled": 480.0,
+    "generated_mean_of_donor_medians": 462.0909118652344,
+    "observed_mean_of_donor_medians": 487.45454545454544
+   },
+   "contrast": {
+    "mean_cp10k": {
+     "per_type": {
+      "T cell": {
+       "r_generated": 0.4955015640545308,
+       "r_broadcast": 0.5426693698646771,
+       "partial_r_generated_given_broadcast": 0.3483993335853783,
+       "r_generated_vs_broadcast": 0.4249364087840753,
+       "n_genes": 2907
+      },
+      "B cell": {
+       "r_generated": 0.23843165332677707,
+       "r_broadcast": 0.46011992006345803,
+       "partial_r_generated_given_broadcast": 0.12854750170213747,
+       "r_generated_vs_broadcast": 0.28007397801020556,
+       "n_genes": 3041
+      },
+      "Monocyte": {
+       "r_generated": 0.3557163348952947,
+       "r_broadcast": 0.5915210794160278,
+       "partial_r_generated_given_broadcast": 0.1550721771974655,
+       "r_generated_vs_broadcast": 0.4084159697412405,
+       "n_genes": 3060
+      },
+      "NK cell": {
+       "r_generated": 0.37319733443777137,
+       "r_broadcast": 0.45003670363580767,
+       "partial_r_generated_given_broadcast": 0.24635767530306976,
+       "r_generated_vs_broadcast": 0.37635295569640487,
+       "n_genes": 2949
+      }
+     },
+     "mean_r_generated": 0.36571172167859345,
+     "mean_r_broadcast": 0.5110867682449927,
+     "mean_partial_r": 0.21959417194701275
+    },
+    "pseudobulk_log1p_cpm": {
+     "per_type": {
+      "T cell": {
+       "r_generated": 0.39932987372898515,
+       "r_broadcast": 0.7425337537971076,
+       "partial_r_generated_given_broadcast": 0.15152247412519773,
+       "r_generated_vs_broadcast": 0.41333363600083306,
+       "n_genes": 2907
+      },
+      "B cell": {
+       "r_generated": 0.16560878455733485,
+       "r_broadcast": 0.6652140288020838,
+       "partial_r_generated_given_broadcast": 0.020652986462832576,
+       "r_generated_vs_broadcast": 0.22637600606689282,
+       "n_genes": 3041
+      },
+      "Monocyte": {
+       "r_generated": 0.2402243450115651,
+       "r_broadcast": 0.7553162442814422,
+       "partial_r_generated_given_broadcast": 0.03987684856844198,
+       "r_generated_vs_broadcast": 0.28487871581989443,
+       "n_genes": 3060
+      },
+      "NK cell": {
+       "r_generated": 0.232791516548745,
+       "r_broadcast": 0.6686724823885001,
+       "partial_r_generated_given_broadcast": 0.045479831025545675,
+       "r_generated_vs_broadcast": 0.29989446409205933,
+       "n_genes": 2949
+      }
+     },
+     "mean_r_generated": 0.25948862996165756,
+     "mean_r_broadcast": 0.7079341273172834,
+     "mean_partial_r": 0.0643830350455045
+    }
+   },
+   "deg_top50": {
+    "types_scored": [
+     "T cell",
+     "B cell",
+     "Monocyte",
+     "NK cell",
+     "DC"
+    ],
+    "types_not_scored": {},
+    "per_type": {
+     "T cell": {
+      "deg_scored": true,
+      "top50_overlap": 2,
+      "top50_jaccard": 0.05,
+      "obs_top_list_size": 39,
+      "gen_top_list_size": 3,
+      "top50_overlap_broadcast": 5,
+      "top50_jaccard_broadcast": 0.12195121951219512,
+      "n_cells_obs": 69840,
+      "n_cells_gen": 9625
+     },
+     "B 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": 0.2631578947368421,
+      "n_cells_obs": 11400,
+      "n_cells_gen": 650
+     },
+     "Monocyte": {
+      "deg_scored": true,
+      "top50_overlap": 0,
+      "top50_jaccard": 0.0,
+      "obs_top_list_size": 6,
+      "gen_top_list_size": 0,
+      "top50_overlap_broadcast": 5,
+      "top50_jaccard_broadcast": 0.625,
+      "n_cells_obs": 18916,
+      "n_cells_gen": 3890
+     },
+     "NK cell": {
+      "deg_scored": true,
+      "top50_overlap": 1,
+      "top50_jaccard": 0.1111111111111111,
+      "obs_top_list_size": 9,
+      "gen_top_list_size": 1,
+      "top50_overlap_broadcast": 5,
+      "top50_jaccard_broadcast": 0.45454545454545453,
+      "n_cells_obs": 12021,
+      "n_cells_gen": 1769
+     },
+     "DC": {
+      "deg_scored": true,
+      "top50_overlap": 0,
+      "top50_jaccard": 0.0,
+      "obs_top_list_size": 4,
+      "gen_top_list_size": 0,
+      "top50_overlap_broadcast": 4,
+      "top50_jaccard_broadcast": 0.5714285714285714,
+      "n_cells_obs": 1706,
+      "n_cells_gen": 68
+     }
+    },
+    "sum": 3,
+    "sum_4type": 3,
+    "max": 75,
+    "mean": 0.6,
+    "sum_broadcast": 24,
+    "sum_broadcast_4type": 20,
+    "broadcast_n_significant": 7,
+    "seconds": 0.02
+   },
+   "seconds": 4.6
+  }
+ },
+ "code_sha256": {
+  "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/tools/score_ft.py": "492e929e9bc7b716f6f9c8bbeedc98e7dec8cc759a20cdc6ec39b1e65194eb74",
+  "/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": 5.1
+}