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/draw4_scsemi5/README.md

@@ -0,0 +1,73 @@
+# score_ft.py output: giroux2022, 10 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 10 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) (1500 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 0.9993, 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
+   0.9987, 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 = 5) 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 5 + 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/draw4_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,2892,0.5541459115565288,0.7550828384073526,0.1622193851692433,0.6237988754803171,5,5,5,5,55877,8516,,,,,,,,,,,
+scsemi5,T cell,pseudobulk_log1p_cpm,2892,0.3682024208693222,0.7802455960807625,-0.023086683209985696,0.4880590142193319,5,5,5,5,55877,8516,,,,,,,,,,,
+scsemi5,B cell,mean_cp10k,3083,0.35362319212602733,0.5987707458214633,0.1558394314016647,0.3994857352086033,5,5,5,5,4890,835,,,,,,,,,,,
+scsemi5,B cell,pseudobulk_log1p_cpm,3083,0.14276558103194806,0.6078954142745557,-0.02617868441231065,0.2677972004680074,5,5,5,5,4890,835,,,,,,,,,,,
+scsemi5,Monocyte,mean_cp10k,3040,0.5075108592374175,0.5783499332800901,0.37265984792978263,0.3944925601550915,5,5,5,5,21294,3057,,,,,,,,,,,
+scsemi5,Monocyte,pseudobulk_log1p_cpm,3040,0.2267245950963786,0.7454060515768156,0.10801304261290195,0.20971537042494268,5,5,5,5,21294,3057,,,,,,,,,,,
+scsemi5,NK cell,mean_cp10k,2963,0.4090252288829422,0.6789110818845145,0.18318123306990283,0.42296032042202175,5,5,5,5,12579,2252,,,,,,,,,,,
+scsemi5,NK cell,pseudobulk_log1p_cpm,2963,0.22466242259134594,0.6850741039839333,0.03361730092945203,0.29376919860715234,5,5,5,5,12579,2252,,,,,,,,,,,
+scsemi5,T cell,welch_top50_deg,5563,,,,,5,5,5,5,55877,8516,True,0,0.0,12,7,12,7,6,0.3333333333333333,12,12
+scsemi5,B cell,welch_top50_deg,5563,,,,,5,5,5,5,4890,835,True,0,0.0,8,0,8,0,6,0.42857142857142855,12,12
+scsemi5,Monocyte,welch_top50_deg,5563,,,,,5,5,5,5,21294,3057,True,0,0.0,12,0,12,0,7,0.4117647058823529,12,12
+scsemi5,NK cell,welch_top50_deg,5563,,,,,5,5,5,5,12579,2252,True,0,0.0,17,0,17,0,6,0.2608695652173913,12,12
+scsemi5,DC,welch_top50_deg,5563,,,,,5,5,5,5,1773,126,True,0,0.0,3,0,3,0,3,0.25,12,12

paper-releases/pbisc-diffusion/updates/20260930T065633Z-summary-deck-scsemi-final/diffusion/dataset/server_1/model8/plans/paired_finetune_intradataset_20260927/scores/giroux2022/fewshot_n5/draw4_scsemi5/per_donor.csv

@@ -0,0 +1,11 @@
+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
+scsemi5,450905_d0,mild,1500,5914,150,0.542,0.30943523841731485,0.08133333333333333,0.036185322962461956,0.184,0.4293202570172472,0.18533333333333332,0.12089956036523504,0.006666666666666667,0.009976327358809604,0.0006666666666666666,0.09418329387893135,0.6842930898433096,0.5987967413223616,0.6145248746216747,2,False,1.0,0.5933333333333334,0.38,0.2733333333333333,18.4,42.93202570172472,484.0,574.5
+scsemi5,450905_d14,mild,1500,7407,150,0.4826666666666667,0.4422843256379101,0.03333333333333333,0.02929661131362225,0.24266666666666667,0.27420008100445525,0.20866666666666667,0.15782368030241664,0.009333333333333334,0.02983664101525584,0.023333333333333334,0.06655866072633995,0.19052409882543542,0.33063170582174406,0.5868049714418707,2,False,0.98,0.5466666666666666,0.47333333333333333,0.26666666666666666,24.266666666666666,27.420008100445525,483.0,558.0
+scsemi5,7085CA_d0,mild,1500,12585,150,0.54,0.21660707191100517,0.066,0.024791418355184745,0.21866666666666668,0.5874453714739769,0.16933333333333334,0.13198251887167262,0.006,0.017719507350019866,0.0,0.021454112038140644,0.8039046483909417,0.8949626862138647,0.5911542159258689,1,True,1.0,0.54,0.4,0.24,21.866666666666667,58.74453714739769,484.0,640.0
+scsemi5,7085CA_d7,mild,1500,9515,150,0.5333333333333333,0.521492380451918,0.03133333333333333,0.0722017866526537,0.20533333333333334,0.10110352075669994,0.20733333333333334,0.2725170782974251,0.008,0.019863373620599055,0.014666666666666666,0.012821860220704152,0.2358311438080224,0.4348078856615509,0.6212050117887834,1,True,0.9933333333333333,0.72,0.5066666666666667,0.24,20.533333333333335,10.110352075669994,484.0,502.0
+scsemi5,7085CA_d14,mild,1500,11887,150,0.538,0.472533019264743,0.034,0.07133843694792631,0.21933333333333332,0.16210986792294102,0.18533333333333332,0.24270211155043325,0.009333333333333334,0.018591738874400604,0.014,0.032724825439555814,0.24538089229129867,0.3709079832657155,0.6355327120667343,1,True,1.0,0.6266666666666667,0.56,0.3,21.933333333333334,16.2109867922941,484.0,536.0
+scsemi5,B85D75_d0,severe,1500,11441,150,0.546,0.7173324010138974,0.016,0.031203566121842496,0.304,0.11441307578008915,0.10933333333333334,0.06039681845992483,0.017333333333333333,0.016519535005681322,0.007333333333333333,0.06013460361856481,0.47867447484194264,0.4453152798334465,0.650891850800176,7,False,1.0,0.6666666666666666,0.6666666666666666,0.13333333333333333,30.4,11.441307578008916,484.0,347.0
+scsemi5,B85D75_d7,severe,1500,12328,150,0.6673333333333333,0.6435756002595717,0.06666666666666667,0.06448734587929916,0.16733333333333333,0.13741077222582737,0.07133333333333333,0.04201817001946788,0.005333333333333333,0.027741726151849447,0.022,0.08476638546398442,0.17034955656500106,0.4045481459575713,0.631437438049459,1,True,1.0,0.74,0.5066666666666667,0.38666666666666666,16.733333333333334,13.741077222582737,484.0,363.0
+scsemi5,180E1A_d0,severe,1500,7157,150,0.6233333333333333,0.7019700992035769,0.09466666666666666,0.04135811094033813,0.16266666666666665,0.13175911694844208,0.10133333333333333,0.0655302501047925,0.004,0.013273718038284197,0.014,0.04610870476456616,0.24003837734618788,0.4544373307053366,0.6154757538794283,8,False,0.9866666666666667,0.6533333333333333,0.5266666666666666,0.20666666666666667,16.266666666666666,13.175911694844208,483.0,433.0
+scsemi5,180E1A_d7,severe,1500,10591,150,0.5413333333333333,0.7348692285903126,0.032,0.04909829100179398,0.21133333333333335,0.09064299877254273,0.184,0.08280615617033331,0.011333333333333334,0.010480596733075253,0.02,0.03210272873194221,0.4454738299814308,0.5242880319872557,0.6130231298946427,5,False,1.0,0.68,0.38666666666666666,0.16666666666666666,21.133333333333333,9.064299877254273,484.0,408.0
+scsemi5,180E1A_d14,severe,1500,12511,150,0.6633333333333333,0.680840860043162,0.10133333333333333,0.05147470226200943,0.12266666666666666,0.12277196067460634,0.07933333333333334,0.0800095915594277,0.006666666666666667,0.009831348413396212,0.02666666666666667,0.05507153704739829,0.09971726214264781,0.424657551112119,0.6645575715478664,1,True,1.0,0.5466666666666666,0.4866666666666667,0.24666666666666667,12.266666666666666,12.277196067460634,484.0,342.0

paper-releases/pbisc-diffusion/updates/20260930T065633Z-summary-deck-scsemi-final/diffusion/dataset/server_1/model8/plans/paired_finetune_intradataset_20260927/scores/giroux2022/fewshot_n5/draw4_scsemi5/summary.json

@@ -0,0 +1,475 @@
+{
+ "schema": "paired_ft_scores.v1",
+ "dataset": "giroux2022",
+ "written_utc": "2026-09-29T23:59:21.182539+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/draw4_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/draw4/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/draw4/scsemi5__raw.npz"
+ ],
+ "arms": [
+  "scsemi5"
+ ],
+ "donors": [
+  "450905_d0",
+  "450905_d14",
+  "7085CA_d0",
+  "7085CA_d7",
+  "7085CA_d14",
+  "B85D75_d0",
+  "B85D75_d7",
+  "180E1A_d0",
+  "180E1A_d7",
+  "180E1A_d14"
+ ],
+ "n_donors": 10,
+ "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": 1500,
+  "pca": {
+   "n_components": 50,
+   "explained_variance_ratio_sum": 0.2558751106262207,
+   "n_fit_cells": 1500
+  },
+  "r2_control": {
+   "precision": 0.9993333333333333,
+   "coverage": 1.0,
+   "density": 6.356199999999999,
+   "n_query": 1500,
+   "depth_matched": {
+    "precision": 0.9986666666666667,
+    "coverage": 1.0,
+    "density": 6.2948
+   }
+  },
+  "observed_l1_loo_null_mean": 0.4883353341880966
+ },
+ "contrast_definition": {
+  "enabled": true,
+  "rule": "explicit --case-groups / --control-groups",
+  "group_source": "features groups",
+  "case_groups": [
+   "severe"
+  ],
+  "control_groups": [
+   "mild"
+  ],
+  "n_case_donors": 5,
+  "n_control_donors": 5,
+  "n_donors_in_neither": 0,
+  "case": [
+   false,
+   false,
+   false,
+   false,
+   false,
+   true,
+   true,
+   true,
+   true,
+   true
+  ],
+  "ctrl": [
+   true,
+   true,
+   true,
+   true,
+   true,
+   false,
+   false,
+   false,
+   false,
+   false
+  ],
+  "groups": [
+   "mild",
+   "mild",
+   "mild",
+   "mild",
+   "mild",
+   "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": 12,
+    "n_tested": 5239,
+    "n_case_donors": 5,
+    "n_control_donors": 5,
+    "n_cells": 55877,
+    "top_list_size": 12
+   },
+   "B cell": {
+    "n_significant": 8,
+    "n_tested": 4859,
+    "n_case_donors": 5,
+    "n_control_donors": 5,
+    "n_cells": 4890,
+    "top_list_size": 8
+   },
+   "Monocyte": {
+    "n_significant": 12,
+    "n_tested": 5165,
+    "n_case_donors": 5,
+    "n_control_donors": 5,
+    "n_cells": 21294,
+    "top_list_size": 12
+   },
+   "NK cell": {
+    "n_significant": 17,
+    "n_tested": 4983,
+    "n_case_donors": 5,
+    "n_control_donors": 5,
+    "n_cells": 12579,
+    "top_list_size": 17
+   },
+   "DC": {
+    "n_significant": 3,
+    "n_tested": 4684,
+    "n_case_donors": 5,
+    "n_control_donors": 5,
+    "n_cells": 1773,
+    "top_list_size": 3
+   }
+  },
+  "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": 0.3594187374036218,
+   "comp_l1_obs_loo_null": 0.4883353341880966,
+   "precision": 0.996,
+   "coverage": 0.6313333333333333,
+   "precision_dm": 0.48933333333333334,
+   "coverage_dm": 0.246,
+   "pb_r_median": 0.6183403828341059,
+   "pb_top1": 0.5,
+   "mono_pct_gen": 20.38,
+   "mono_pct_obs": 21.511770225768277,
+   "median_umi_gen": 484.0,
+   "median_umi_obs": 447.0,
+   "contrast_mean_cp10k_r_gen": 0.4560762979507289,
+   "contrast_pseudobulk_log1p_cpm_r_gen": 0.2405887548972487,
+   "contrast_mean_cp10k_r_broadcast": 0.6527786498483552,
+   "contrast_pseudobulk_log1p_cpm_r_broadcast": 0.7046552914790167,
+   "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": 52,
+   "deg_top50_sum_4type": 0,
+   "deg_top50_sum_broadcast": 28,
+   "seconds": 4.4
+  }
+ },
+ "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/draw4/scsemi5__raw.npz",
+   "npz_sha256": "93e63e4411e90a2eb272b7b73101a5015edce263e950fadedcf01d5b6e481634",
+   "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/draw4/test_raw.npz",
+    "input_condition_sha256": "1339e7631cd54c77bef770b2248bdc2ca560a6dc70e244ed61cd4d38a835bf3f",
+    "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/draw4/test_raw.npz",
+    "sha256": "1339e7631cd54c77bef770b2248bdc2ca560a6dc70e244ed61cd4d38a835bf3f"
+   },
+   "poisson": {
+    "seed": 20260916,
+    "rule": "whole NPZ, rows in NPZ order",
+    "nnz_whole_npz": 1687839,
+    "median_library_whole_npz_panel": 484.0,
+    "n_rows_whole_npz": 15000
+   },
+   "annotation": {
+    "model_pt_sha256": "bf3655899e5abee435340fc008656c4a17342da0f30b70cfd43de8e9223e1930",
+    "zero_padded_model_genes": 0,
+    "label_counts": {
+     "T cell": 8516,
+     "Monocyte": 3057,
+     "NK cell": 2252,
+     "B cell": 835,
+     "Other": 214,
+     "DC": 126
+    },
+    "mean_l2_max_prob": 0.8520268356353045,
+    "predict_seconds": 2.9
+   },
+   "n_donors": 10,
+   "n_generated_cells": 15000,
+   "composition": {
+    "l1_mean": 0.3594187374036218,
+    "l1_median": 0.24270963481874328,
+    "l1_obs_loo_null_mean": 0.4883353341880966,
+    "l1_generated_loo_mean_vs_observed_mean": 0.43165114867800003,
+    "composition_top1": 0.2,
+    "mean_fraction_generated": {
+     "T cell": 0.567733,
+     "B cell": 0.055667,
+     "Monocyte": 0.2038,
+     "NK cell": 0.150133,
+     "DC": 0.0084,
+     "Other": 0.014267
+    },
+    "mean_fraction_observed": {
+     "T cell": 0.544094,
+     "B cell": 0.047144,
+     "Monocyte": 0.215118,
+     "NK cell": 0.125669,
+     "DC": 0.017383,
+     "Other": 0.050593
+    }
+   },
+   "pseudobulk": {
+    "n_genes": 5563,
+    "r_median": 0.6183403828341059,
+    "r_mean": 0.6224607530016504,
+    "top1": 0.5,
+    "n_top1": 5,
+    "own_rank_median": 1.5,
+    "chance_top1": 0.1
+   },
+   "precision_coverage": {
+    "precision": 0.996,
+    "coverage": 0.6313333333333333,
+    "density": 1.2738666666666667,
+    "n_query": 1500,
+    "n_reference": 1500
+   },
+   "precision_coverage_depth_matched": {
+    "precision": 0.48933333333333334,
+    "coverage": 0.246,
+    "density": 0.18233333333333332,
+    "detected_median_before": {
+     "R1": 252.0,
+     "Q": 111.0
+    },
+    "detected_median_after": {
+     "R1": 111.0,
+     "Q": 111.0
+    }
+   },
+   "monocyte_pct": {
+    "generated_mean_over_donors": 20.38,
+    "observed_mean_over_donors": 21.511770225768277,
+    "generated_pooled_cells": 20.380000000000003
+   },
+   "median_umi_per_cell_scoring_genes": {
+    "generated_pooled": 484.0,
+    "observed_pooled": 447.0,
+    "generated_mean_of_donor_medians": 483.79998779296875,
+    "observed_mean_of_donor_medians": 470.35
+   },
+   "contrast": {
+    "mean_cp10k": {
+     "per_type": {
+      "T cell": {
+       "r_generated": 0.5541459115565288,
+       "r_broadcast": 0.7550828384073526,
+       "partial_r_generated_given_broadcast": 0.1622193851692433,
+       "r_generated_vs_broadcast": 0.6237988754803171,
+       "n_genes": 2892
+      },
+      "B cell": {
+       "r_generated": 0.35362319212602733,
+       "r_broadcast": 0.5987707458214633,
+       "partial_r_generated_given_broadcast": 0.1558394314016647,
+       "r_generated_vs_broadcast": 0.3994857352086033,
+       "n_genes": 3083
+      },
+      "Monocyte": {
+       "r_generated": 0.5075108592374175,
+       "r_broadcast": 0.5783499332800901,
+       "partial_r_generated_given_broadcast": 0.37265984792978263,
+       "r_generated_vs_broadcast": 0.3944925601550915,
+       "n_genes": 3040
+      },
+      "NK cell": {
+       "r_generated": 0.4090252288829422,
+       "r_broadcast": 0.6789110818845145,
+       "partial_r_generated_given_broadcast": 0.18318123306990283,
+       "r_generated_vs_broadcast": 0.42296032042202175,
+       "n_genes": 2963
+      }
+     },
+     "mean_r_generated": 0.4560762979507289,
+     "mean_r_broadcast": 0.6527786498483552,
+     "mean_partial_r": 0.21847497439264837
+    },
+    "pseudobulk_log1p_cpm": {
+     "per_type": {
+      "T cell": {
+       "r_generated": 0.3682024208693222,
+       "r_broadcast": 0.7802455960807625,
+       "partial_r_generated_given_broadcast": -0.023086683209985696,
+       "r_generated_vs_broadcast": 0.4880590142193319,
+       "n_genes": 2892
+      },
+      "B cell": {
+       "r_generated": 0.14276558103194806,
+       "r_broadcast": 0.6078954142745557,
+       "partial_r_generated_given_broadcast": -0.02617868441231065,
+       "r_generated_vs_broadcast": 0.2677972004680074,
+       "n_genes": 3083
+      },
+      "Monocyte": {
+       "r_generated": 0.2267245950963786,
+       "r_broadcast": 0.7454060515768156,
+       "partial_r_generated_given_broadcast": 0.10801304261290195,
+       "r_generated_vs_broadcast": 0.20971537042494268,
+       "n_genes": 3040
+      },
+      "NK cell": {
+       "r_generated": 0.22466242259134594,
+       "r_broadcast": 0.6850741039839333,
+       "partial_r_generated_given_broadcast": 0.03361730092945203,
+       "r_generated_vs_broadcast": 0.29376919860715234,
+       "n_genes": 2963
+      }
+     },
+     "mean_r_generated": 0.2405887548972487,
+     "mean_r_broadcast": 0.7046552914790167,
+     "mean_partial_r": 0.02309124398001441
+    }
+   },
+   "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": 12,
+      "gen_top_list_size": 7,
+      "top50_overlap_broadcast": 6,
+      "top50_jaccard_broadcast": 0.3333333333333333,
+      "n_cells_obs": 55877,
+      "n_cells_gen": 8516
+     },
+     "B cell": {
+      "deg_scored": true,
+      "top50_overlap": 0,
+      "top50_jaccard": 0.0,
+      "obs_top_list_size": 8,
+      "gen_top_list_size": 0,
+      "top50_overlap_broadcast": 6,
+      "top50_jaccard_broadcast": 0.42857142857142855,
+      "n_cells_obs": 4890,
+      "n_cells_gen": 835
+     },
+     "Monocyte": {
+      "deg_scored": true,
+      "top50_overlap": 0,
+      "top50_jaccard": 0.0,
+      "obs_top_list_size": 12,
+      "gen_top_list_size": 0,
+      "top50_overlap_broadcast": 7,
+      "top50_jaccard_broadcast": 0.4117647058823529,
+      "n_cells_obs": 21294,
+      "n_cells_gen": 3057
+     },
+     "NK cell": {
+      "deg_scored": true,
+      "top50_overlap": 0,
+      "top50_jaccard": 0.0,
+      "obs_top_list_size": 17,
+      "gen_top_list_size": 0,
+      "top50_overlap_broadcast": 6,
+      "top50_jaccard_broadcast": 0.2608695652173913,
+      "n_cells_obs": 12579,
+      "n_cells_gen": 2252
+     },
+     "DC": {
+      "deg_scored": true,
+      "top50_overlap": 0,
+      "top50_jaccard": 0.0,
+      "obs_top_list_size": 3,
+      "gen_top_list_size": 0,
+      "top50_overlap_broadcast": 3,
+      "top50_jaccard_broadcast": 0.25,
+      "n_cells_obs": 1773,
+      "n_cells_gen": 126
+     }
+    },
+    "sum": 0,
+    "sum_4type": 0,
+    "max": 52,
+    "mean": 0.0,
+    "sum_broadcast": 28,
+    "sum_broadcast_4type": 25,
+    "broadcast_n_significant": 12,
+    "seconds": 0.02
+   },
+   "seconds": 4.4
+  }
+ },
+ "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": 5.0
+}