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/ji2025_chili/fewshot_n5/draw2_scsemi5/README.md

@@ -0,0 +1,73 @@
+# score_ft.py output: ji2025_chili, 26 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/ji2025_chili/reference.npz` (sha 19c30f56db69), built from datadir:/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data/ji2025_chili;
+  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/ji2025_chili/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 (1004 genes).
+- Scoring genes: 6k panel genes measured by both assays in every sample of the dataset (5752 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 26 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) (3900 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 0.9969, 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
+   0.9972, 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 ['TOL', 'FU1'] (n = 16) vs control groups ['BC', 'AC'] (n = 10); 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 5752 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/ji2025_chili/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 16 + 10
+   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/ji2025_chili/fewshot_n5/draw2_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,3072,0.2098165747465703,0.2701865645650968,0.12622048291840757,0.35629362939716336,16,10,16,10,42751,21936,,,,,,,,,,,
+scsemi5,T cell,pseudobulk_log1p_cpm,3072,0.10685974295234307,0.29511657864157254,0.054508510436193425,0.1887916056253993,16,10,16,10,42751,21936,,,,,,,,,,,
+scsemi5,B cell,mean_cp10k,3277,0.18457534118803495,0.15572870002588607,0.14460913208606,0.3145193191198802,16,10,14,9,5672,688,,,,,,,,,,,
+scsemi5,B cell,pseudobulk_log1p_cpm,3277,-0.014817159195365519,0.06708509712458424,-0.023863605332527807,0.13099082120335118,16,10,14,9,5672,688,,,,,,,,,,,
+scsemi5,Monocyte,mean_cp10k,3152,0.18892332636666287,0.37567927431475395,0.058071878919912126,0.369783409371101,16,10,16,9,28874,8190,,,,,,,,,,,
+scsemi5,Monocyte,pseudobulk_log1p_cpm,3152,0.0882137599243738,0.33780407869585316,0.023774710288588918,0.19618312707273142,16,10,16,9,28874,8190,,,,,,,,,,,
+scsemi5,NK cell,mean_cp10k,3069,0.14299134222648124,0.2533010131298753,0.06018363378366271,0.3491259174809879,16,10,16,9,14784,5511,,,,,,,,,,,
+scsemi5,NK cell,pseudobulk_log1p_cpm,3069,0.0959622271335853,0.1357496033139099,0.06976488289472567,0.20898288022567643,16,10,16,9,14784,5511,,,,,,,,,,,
+scsemi5,T cell,welch_top50_deg,5752,,,,,16,10,16,10,42751,21936,True,0,0.0,50,0,2050,0,4,0.05714285714285714,24,24
+scsemi5,B cell,welch_top50_deg,5752,,,,,16,10,14,9,5672,688,True,0,0.0,50,0,94,0,1,0.0136986301369863,24,24
+scsemi5,Monocyte,welch_top50_deg,5752,,,,,16,10,16,9,28874,8190,True,0,0.0,50,0,266,0,0,0.0,24,24
+scsemi5,NK cell,welch_top50_deg,5752,,,,,16,10,16,9,14784,5511,True,0,0.0,50,0,397,0,0,0.0,24,24
+scsemi5,DC,welch_top50_deg,5752,,,,,14,10,14,9,1112,1223,True,0,0.0,50,0,58,0,0,0.0,24,24

paper-releases/pbisc-diffusion/updates/20260930T065633Z-summary-deck-scsemi-final/diffusion/dataset/server_1/model8/plans/paired_finetune_intradataset_20260927/scores/ji2025_chili/fewshot_n5/draw2_scsemi5/per_donor.csv

@@ -0,0 +1,27 @@
+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,N064_TOL,TOL,1500,2498,150,0.8053333333333333,0.6517213771016813,0.016666666666666666,0.044035228182546036,0.078,0.1677341873498799,0.08533333333333333,0.07606084867894315,0.011333333333333334,0.010008006405124099,0.0033333333333333335,0.05044035228182546,0.32841953562850285,0.44066885632062114,0.5938115902129082,24,False,0.94,0.24,0.25333333333333335,0.11333333333333333,7.8,16.77341873498799,854.0,662.0
+scsemi5,N064_FU1,FU1,1500,4659,150,0.82,0.6396222365314446,0.015333333333333332,0.016527151749302426,0.07466666666666667,0.13865636402661516,0.07666666666666666,0.19167203262502683,0.012,0.00472204335694355,0.0013333333333333333,0.008800171710667526,0.37531144022322377,0.447147467447507,0.5964374367802698,16,False,0.9333333333333333,0.04,0.31333333333333335,0.02,7.466666666666668,13.865636402661515,852.0,925.0
+scsemi5,N038_BC,BC,1500,2801,150,0.36,0.5608711174580507,0.020666666666666667,0.0046411995715815784,0.2813333333333333,0.16565512317029632,0.27066666666666667,0.2170653338093538,0.050666666666666665,0.034273473759371655,0.016666666666666666,0.017493752231345947,0.40339640604546,0.3913209350209281,0.5875578814809749,10,False,0.98,0.07333333333333333,0.4266666666666667,0.20666666666666667,28.133333333333333,16.565512317029633,854.0,1317.0
+scsemi5,N038_AC,AC,1500,3976,150,0.488,0.6830985915492958,0.014666666666666666,0.00829979879275654,0.31466666666666665,0.1290241448692153,0.14066666666666666,0.12072434607645875,0.012,0.03194164989939638,0.03,0.026911468812877263,0.43008048289738426,0.5093694384341356,0.5669785845886196,5,False,0.9533333333333334,0.23333333333333334,0.5,0.19333333333333333,31.466666666666665,12.90241448692153,855.0,910.5
+scsemi5,N055_AC,AC,1500,4308,150,0.6926666666666667,0.340297121634169,0.05533333333333333,0.07799442896935933,0.11666666666666667,0.32474466109563604,0.09866666666666667,0.2051996285979573,0.006,0.0060352831940575676,0.030666666666666665,0.0457288765088208,0.7047390900649952,0.2683259872552967,0.6325476593459649,4,False,0.82,0.1,0.36,0.23333333333333334,11.666666666666666,32.47446610956361,852.5,1139.5
+scsemi5,N136_TOL,TOL,1500,759,150,0.8766666666666667,0.48616600790513836,0.005333333333333333,0.10013175230566534,0.08066666666666666,0.19104084321475626,0.010666666666666666,0.0922266139657444,0.014666666666666666,0.0013175230566534915,0.012,0.12911725955204217,0.807699604743083,0.3440190413647225,0.4548222265932968,10,False,0.8333333333333334,0.02666666666666667,0.34,0.07333333333333333,8.066666666666666,19.104084321475625,853.0,1837.0
+scsemi5,N136_FU1,FU1,1500,2927,150,0.8633333333333333,0.4246668944311582,0.0,0.14724974376494704,0.07866666666666666,0.19337205329689103,0.0013333333333333333,0.09531943969935087,0.0,0.0023915271609156134,0.056666666666666664,0.13700034164673727,0.8773328778043503,0.410613532351375,0.4442168426946195,3,False,0.8533333333333334,0.08,0.5933333333333334,0.17333333333333334,7.866666666666666,19.3372053296891,854.0,1501.0
+scsemi5,N138_TOL,TOL,1500,5129,150,0.3446666666666667,0.13803860401637746,0.022,0.02807564827451745,0.29333333333333333,0.5878338857477091,0.252,0.17703255995320724,0.05466666666666667,0.006044063170208618,0.03333333333333333,0.06297523883798012,0.6604362123870801,0.7733211231673002,0.5672929179844587,2,False,0.62,0.19333333333333333,0.62,0.09333333333333334,29.333333333333332,58.78338857477091,854.5,1117.0
+scsemi5,N138_FU1,FU1,1500,2107,150,0.724,0.2728998576174656,0.05533333333333333,0.04841006169909824,0.09866666666666667,0.4831514000949217,0.092,0.1134314190792596,0.016,0.005220692928334125,0.014,0.07688656858092074,0.9376054421768707,0.5540453704150132,0.5883074953949926,10,False,0.5733333333333334,0.11333333333333333,0.2733333333333333,0.21333333333333335,9.866666666666667,48.31514000949217,854.0,1446.0
+scsemi5,N128_BC,BC,1500,3902,150,0.36533333333333334,0.5158892875448488,0.019333333333333334,0.06817016914402871,0.2773333333333333,0.29420809841107126,0.174,0.07842132239876987,0.044,0.03331624807790876,0.12,0.00999487442337263,0.43253511019989743,0.22126235566025862,0.5502653961968691,4,False,0.9133333333333333,0.24,0.37333333333333335,0.16,27.73333333333333,29.420809841107125,854.0,1088.0
+scsemi5,N128_AC,AC,1500,2713,150,0.3586666666666667,0.5488389237007003,0.028666666666666667,0.06524143015112421,0.2886666666666667,0.2790269074824917,0.20933333333333334,0.07077036490969407,0.059333333333333335,0.029487652045705862,0.05533333333333333,0.006634721710283818,0.45349404103698243,0.24328457606458784,0.5388630503446206,13,False,0.9333333333333333,0.15333333333333332,0.4066666666666667,0.14666666666666667,28.866666666666667,27.90269074824917,853.0,1181.0
+scsemi5,N129_BC,BC,1500,3087,150,0.464,0.5963718820861678,0.0013333333333333333,0.11305474570780694,0.3,0.08357628765792031,0.11066666666666666,0.18140589569160998,0.034666666666666665,0.013281503077421444,0.08933333333333333,0.012309685779073534,0.6296650469711694,0.4599645175272137,0.48724308468122546,4,False,0.5666666666666667,0.22666666666666666,0.38,0.12,30.0,8.35762876579203,854.0,1059.0
+scsemi5,N129_AC,AC,1500,2798,150,0.37333333333333335,0.7076483202287348,0.02266666666666667,0.04932094353109364,0.2693333333333333,0.07969978556111508,0.238,0.13366690493209435,0.06866666666666667,0.0228734810578985,0.028,0.006790564689063617,0.7219385275196569,0.5428317542586936,0.5462854814053026,8,False,0.92,0.3333333333333333,0.44,0.23333333333333334,26.93333333333333,7.969978556111508,854.0,759.0
+scsemi5,N150_BC,BC,1500,2433,150,0.43533333333333335,0.802301685162351,0.010666666666666666,0.02301685162351007,0.312,0.1385121249486231,0.19,0.02178380600082203,0.014,0.005754212905877518,0.038,0.008631319358816275,0.7586370735717222,0.7041534876493698,0.5858909641734336,4,False,0.98,0.02666666666666667,0.5333333333333333,0.26,31.2,13.851212494862311,854.0,1629.0
+scsemi5,N158_AC,AC,1500,3749,150,0.4726666666666667,0.24619898639637236,0.0006666666666666666,0.026673779674579887,0.38666666666666666,0.6295012003200854,0.05266666666666667,0.04934649239797279,0.014666666666666666,0.024806615097359295,0.07266666666666667,0.0234729261136303,0.5579631901840492,0.7651609044087362,0.5060027691900216,9,False,0.8133333333333334,0.17333333333333334,0.7,0.11333333333333333,38.666666666666664,62.95012003200854,855.0,1047.0
+scsemi5,N159_BC,BC,1500,3779,150,0.98,0.6996559936491136,0.0,0.018523418893887273,0.0,0.08123842286319131,0.0,0.173590897062715,0.0,0.014024874305371792,0.02,0.01296639322572109,0.5747552262503307,0.5299123518003033,0.3854558071169306,1,True,0.04666666666666667,0.14,1.0,0.43333333333333335,0.0,8.123842286319132,855.0,1045.0
+scsemi5,N001_TOL,TOL,1500,5658,150,0.6313333333333333,0.51926475786497,0.0006666666666666666,0.03375751148815836,0.224,0.21067515022976316,0.016,0.2163308589607635,0.09866666666666667,0.003711558854718982,0.029333333333333333,0.016260162601626018,0.4668434075645105,0.223121184837707,0.46514099968321887,19,False,0.88,0.09333333333333334,0.6666666666666666,0.0,22.400000000000002,21.067515022976316,855.0,787.5
+scsemi5,N001_FU1,FU1,1500,4547,150,0.4186666666666667,0.29601935341983726,0.010666666666666666,0.06575764240158347,0.30533333333333335,0.48823400043985044,0.21533333333333332,0.13591378931163403,0.020666666666666667,0.0010996261271167802,0.029333333333333333,0.012975588299978007,0.47598328568286785,0.4496582811729592,0.5486817585852638,14,False,0.8933333333333333,0.08,0.24666666666666667,0.02,30.533333333333335,48.823400043985046,854.0,1165.0
+scsemi5,N002_TOL,TOL,1500,6318,150,0.358,0.41579613801836024,0.027333333333333334,0.05777144666033555,0.22266666666666668,0.3779677113010446,0.25466666666666665,0.1179170623615068,0.114,0.020101297879075657,0.023333333333333334,0.010446343779677113,0.4870705919594808,0.2485688582152307,0.5102413950183972,16,False,0.9533333333333334,0.1,0.38,0.02,22.26666666666667,37.79677113010446,854.0,933.5
+scsemi5,N002_FU1,FU1,1500,4975,150,0.434,0.31618090452261305,0.006666666666666667,0.07276381909547738,0.2653333333333333,0.45587939698492463,0.19733333333333333,0.12261306532663317,0.068,0.02271356783919598,0.028666666666666667,0.00984924623115578,0.513286432160804,0.43672029562758785,0.49982759946396665,21,False,0.9333333333333333,0.04666666666666667,0.49333333333333335,0.006666666666666667,26.53333333333333,45.58793969849246,855.0,979.0
+scsemi5,N012_TOL,TOL,1500,6709,150,0.31866666666666665,0.34759278581010583,0.027333333333333334,0.07884930690117753,0.324,0.4672827545088687,0.24733333333333332,0.08570576837084513,0.038,0.0005962140408406618,0.04466666666666667,0.01997317036816217,0.4474496944403041,0.4550105913719193,0.5825533103259571,16,False,0.86,0.04,0.16666666666666666,0.04,32.4,46.72827545088687,852.0,1193.0
+scsemi5,N012_FU1,FU1,1500,4369,150,0.632,0.35889219501029984,0.056666666666666664,0.19249256122682537,0.11066666666666666,0.3126573586633097,0.09466666666666666,0.10849164568551156,0.006666666666666667,0.000686655985351339,0.09933333333333333,0.02677958342870222,0.7032831311512934,0.373046308332097,0.6010695915508996,20,False,0.68,0.03333333333333333,0.22666666666666666,0.08,11.066666666666666,31.265735866330967,854.0,1299.0
+scsemi5,N022_TOL,TOL,1500,3651,150,0.37933333333333336,0.4864420706655711,0.007333333333333333,0.03697617091207888,0.2846666666666667,0.045740892906053134,0.25733333333333336,0.4182415776499589,0.03333333333333333,0.0,0.038,0.01259928786633799,0.5953196384552177,0.6009038417749315,0.5019215659580503,16,False,0.86,0.02,0.21333333333333335,0.08,28.46666666666667,4.574089290605313,853.0,1410.0
+scsemi5,N022_FU1,FU1,1500,2463,150,0.8133333333333334,0.48477466504263095,0.014,0.048721071863580996,0.08066666666666666,0.20259845716605765,0.07933333333333334,0.25334957369062117,0.01,0.0,0.0026666666666666666,0.010556232237109216,0.6771173365814047,0.26778502261037573,0.6275770536057358,18,False,0.9733333333333334,0.02,0.21333333333333335,0.04666666666666667,8.066666666666666,20.259845716605763,854.0,1459.0
+scsemi5,N043_TOL,TOL,1500,2074,150,0.33066666666666666,0.3548698167791707,0.019333333333333334,0.05737704918032787,0.3006666666666667,0.04580520732883317,0.304,0.4811957569913211,0.013333333333333334,0.0028929604628736743,0.032,0.05785920925747348,0.5306036644165864,0.7271203177048906,0.6300410366498344,19,False,0.9466666666666667,0.006666666666666667,0.20666666666666667,0.08,30.06666666666667,4.580520732883317,854.0,1315.5
+scsemi5,N043_FU1,FU1,1500,3437,150,0.884,0.1638056444573756,0.0,0.06779167878964212,0.09133333333333334,0.5728833284841431,0.005333333333333333,0.14896712249054408,0.0,0.0034914169333721268,0.019333333333333334,0.043060808844922895,1.440388711085249,0.6515440045178634,0.39399330850997727,22,False,0.9,0.04,0.7733333333333333,0.013333333333333334,9.133333333333333,57.28833284841431,854.0,1422.0

paper-releases/pbisc-diffusion/updates/20260930T065633Z-summary-deck-scsemi-final/diffusion/dataset/server_1/model8/plans/paired_finetune_intradataset_20260927/scores/ji2025_chili/fewshot_n5/draw2_scsemi5/summary.json

@@ -0,0 +1,541 @@
+{
+ "schema": "paired_ft_scores.v1",
+ "dataset": "ji2025_chili",
+ "written_utc": "2026-09-29T07:28:33.398636+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",
+  "ji2025_chili",
+  "--out",
+  "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scores/ji2025_chili/fewshot_n5/draw2_scsemi5",
+  "--threads",
+  "6",
+  "--donors",
+  "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/data/ji2025_chili/fewshot_n5/draw2/test_samples.txt",
+  "--case-groups",
+  "TOL,FU1",
+  "--control-groups",
+  "BC,AC",
+  "--arm",
+  "scsemi5=/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/generated/ji2025_chili/fewshot_n5/draw2/scsemi5__raw.npz"
+ ],
+ "arms": [
+  "scsemi5"
+ ],
+ "donors": [
+  "N064_TOL",
+  "N064_FU1",
+  "N038_BC",
+  "N038_AC",
+  "N055_AC",
+  "N136_TOL",
+  "N136_FU1",
+  "N138_TOL",
+  "N138_FU1",
+  "N128_BC",
+  "N128_AC",
+  "N129_BC",
+  "N129_AC",
+  "N150_BC",
+  "N158_AC",
+  "N159_BC",
+  "N001_TOL",
+  "N001_FU1",
+  "N002_TOL",
+  "N002_FU1",
+  "N012_TOL",
+  "N012_FU1",
+  "N022_TOL",
+  "N022_FU1",
+  "N043_TOL",
+  "N043_FU1"
+ ],
+ "n_donors": 26,
+ "reference": {
+  "cache": "/home/sj_server_1/PBISC/immun-pbmc-fm/model_8_sparse_data_diffusion/plans/paired_finetune_intradataset_20260927/scoring_cache/ji2025_chili",
+  "reference_sha256": "19c30f56db699eb4f5a642927c1cd6e173212b877c29ff20d84e60a5ecccef6d",
+  "features_sha256": "049795e3a2470a223347015c0c965f56ca90c21143aad6f72d12e39f3e95e58b",
+  "n_r1_cells": 3900,
+  "pca": {
+   "n_components": 50,
+   "explained_variance_ratio_sum": 0.28848475217819214,
+   "n_fit_cells": 3900
+  },
+  "r2_control": {
+   "precision": 0.9969230769230769,
+   "coverage": 1.0,
+   "density": 2.5067692307692306,
+   "n_query": 3900,
+   "depth_matched": {
+    "precision": 0.9971794871794872,
+    "coverage": 1.0,
+    "density": 2.4707948717948716
+   }
+  },
+  "observed_l1_loo_null_mean": 0.46303386174275474
+ },
+ "contrast_definition": {
+  "enabled": true,
+  "rule": "explicit --case-groups / --control-groups",
+  "group_source": "features groups",
+  "case_groups": [
+   "TOL",
+   "FU1"
+  ],
+  "control_groups": [
+   "BC",
+   "AC"
+  ],
+  "n_case_donors": 16,
+  "n_control_donors": 10,
+  "n_donors_in_neither": 0,
+  "case": [
+   true,
+   true,
+   false,
+   false,
+   false,
+   true,
+   true,
+   true,
+   true,
+   false,
+   false,
+   false,
+   false,
+   false,
+   false,
+   false,
+   true,
+   true,
+   true,
+   true,
+   true,
+   true,
+   true,
+   true,
+   true,
+   true
+  ],
+  "ctrl": [
+   false,
+   false,
+   true,
+   true,
+   true,
+   false,
+   false,
+   false,
+   false,
+   true,
+   true,
+   true,
+   true,
+   true,
+   true,
+   true,
+   false,
+   false,
+   false,
+   false,
+   false,
+   false,
+   false,
+   false,
+   false,
+   false
+  ],
+  "groups": [
+   "TOL",
+   "FU1",
+   "BC",
+   "AC",
+   "AC",
+   "TOL",
+   "FU1",
+   "TOL",
+   "FU1",
+   "BC",
+   "AC",
+   "BC",
+   "AC",
+   "BC",
+   "AC",
+   "BC",
+   "TOL",
+   "FU1",
+   "TOL",
+   "FU1",
+   "TOL",
+   "FU1",
+   "TOL",
+   "FU1",
+   "TOL",
+   "FU1"
+  ]
+ },
+ "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": 5752,
+  "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/ji2025_chili/deg_reference.npz",
+  "aggregates_sha256": "4f5eb8c8389ef1308ce3ec44be8be024cedb97de37351d109d9070c818df5a9e",
+  "aggregates_note": null,
+  "types_not_scored_observed": [],
+  "observed": {
+   "T cell": {
+    "n_significant": 2050,
+    "n_tested": 5676,
+    "n_case_donors": 16,
+    "n_control_donors": 10,
+    "n_cells": 42751,
+    "top_list_size": 50
+   },
+   "B cell": {
+    "n_significant": 94,
+    "n_tested": 5364,
+    "n_case_donors": 16,
+    "n_control_donors": 10,
+    "n_cells": 5672,
+    "top_list_size": 50
+   },
+   "Monocyte": {
+    "n_significant": 266,
+    "n_tested": 5691,
+    "n_case_donors": 16,
+    "n_control_donors": 10,
+    "n_cells": 28874,
+    "top_list_size": 50
+   },
+   "NK cell": {
+    "n_significant": 397,
+    "n_tested": 5482,
+    "n_case_donors": 16,
+    "n_control_donors": 10,
+    "n_cells": 14784,
+    "top_list_size": 50
+   },
+   "DC": {
+    "n_significant": 58,
+    "n_tested": 4980,
+    "n_case_donors": 14,
+    "n_control_donors": 10,
+    "n_cells": 1112,
+    "top_list_size": 50
+   }
+  },
+  "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": 0.6150521231231723,
+   "comp_l1_obs_loo_null": 0.46303386174275474,
+   "precision": 0.8292307692307692,
+   "coverage": 0.1105128205128205,
+   "precision_dm": 0.4269230769230769,
+   "coverage_dm": 0.12205128205128205,
+   "pb_r_median": 0.5494735773910664,
+   "pb_top1": 0.038461538461538464,
+   "mono_pct_gen": 21.0,
+   "mono_pct_obs": 27.487306723551026,
+   "median_umi_gen": 854.0,
+   "median_umi_obs": 1110.0,
+   "contrast_mean_cp10k_r_gen": 0.18157664613193736,
+   "contrast_pseudobulk_log1p_cpm_r_gen": 0.06905464270373417,
+   "contrast_mean_cp10k_r_broadcast": 0.263723888008903,
+   "contrast_pseudobulk_log1p_cpm_r_broadcast": 0.20893883944397995,
+   "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": 250,
+   "deg_top50_sum_4type": 0,
+   "deg_top50_sum_broadcast": 5,
+   "seconds": 7.6
+  }
+ },
+ "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/ji2025_chili/fewshot_n5/draw2/scsemi5__raw.npz",
+   "npz_sha256": "d1f29012e7ad26fe076a61b68b297344ca96d9d9ee0bad2e8dc5639bc0540a7a",
+   "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/ji2025_chili/fewshot_n5/draw2/test_raw.npz",
+    "input_condition_sha256": "70953ef3c751fa832f2c8bf184fa1cd9fc6ef2e04329101441f0fce90b64ddc0",
+    "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/ji2025_chili/fewshot_n5/draw2/test_raw.npz",
+    "sha256": "70953ef3c751fa832f2c8bf184fa1cd9fc6ef2e04329101441f0fce90b64ddc0"
+   },
+   "poisson": {
+    "seed": 20260916,
+    "rule": "whole NPZ, rows in NPZ order",
+    "nnz_whole_npz": 4456877,
+    "median_library_whole_npz_panel": 854.0,
+    "n_rows_whole_npz": 39000
+   },
+   "annotation": {
+    "model_pt_sha256": "bf3655899e5abee435340fc008656c4a17342da0f30b70cfd43de8e9223e1930",
+    "zero_padded_model_genes": 1,
+    "label_counts": {
+     "T cell": 21936,
+     "Monocyte": 8190,
+     "NK cell": 5511,
+     "Other": 1452,
+     "DC": 1223,
+     "B cell": 688
+    },
+    "mean_l2_max_prob": 0.8819687465463694,
+    "predict_seconds": 3.5
+   },
+   "n_donors": 26,
+   "n_generated_cells": 39000,
+   "composition": {
+    "l1_mean": 0.6150521231231723,
+    "l1_median": 0.56635920821719,
+    "l1_obs_loo_null_mean": 0.46303386174275474,
+    "l1_generated_loo_mean_vs_observed_mean": 0.47514122211589704,
+    "composition_top1": 0.0,
+    "mean_fraction_generated": {
+     "T cell": 0.562462,
+     "B cell": 0.017641,
+     "Monocyte": 0.21,
+     "NK cell": 0.141308,
+     "DC": 0.031359,
+     "Other": 0.037231
+    },
+    "mean_fraction_observed": {
+     "T cell": 0.463766,
+     "B cell": 0.059678,
+     "Monocyte": 0.274873,
+     "NK cell": 0.157171,
+     "DC": 0.011569,
+     "Other": 0.032943
+    }
+   },
+   "pseudobulk": {
+    "n_genes": 5752,
+    "r_median": 0.5494735773910664,
+    "r_mean": 0.5381933387513772,
+    "top1": 0.038461538461538464,
+    "n_top1": 1,
+    "own_rank_median": 11.5,
+    "chance_top1": 0.038461538461538464
+   },
+   "precision_coverage": {
+    "precision": 0.8292307692307692,
+    "coverage": 0.1105128205128205,
+    "density": 0.34207692307692306,
+    "n_query": 3900,
+    "n_reference": 3900
+   },
+   "precision_coverage_depth_matched": {
+    "precision": 0.4269230769230769,
+    "coverage": 0.12205128205128205,
+    "density": 0.14064102564102565,
+    "detected_median_before": {
+     "R1": 368.0,
+     "Q": 116.0
+    },
+    "detected_median_after": {
+     "R1": 116.0,
+     "Q": 116.0
+    }
+   },
+   "monocyte_pct": {
+    "generated_mean_over_donors": 21.0,
+    "observed_mean_over_donors": 27.487306723551026,
+    "generated_pooled_cells": 21.0
+   },
+   "median_umi_per_cell_scoring_genes": {
+    "generated_pooled": 854.0,
+    "observed_pooled": 1110.0,
+    "generated_mean_of_donor_medians": 853.8846435546875,
+    "observed_mean_of_donor_medians": 1177.9423076923076
+   },
+   "contrast": {
+    "mean_cp10k": {
+     "per_type": {
+      "T cell": {
+       "r_generated": 0.2098165747465703,
+       "r_broadcast": 0.2701865645650968,
+       "partial_r_generated_given_broadcast": 0.12622048291840757,
+       "r_generated_vs_broadcast": 0.35629362939716336,
+       "n_genes": 3072
+      },
+      "B cell": {
+       "r_generated": 0.18457534118803495,
+       "r_broadcast": 0.15572870002588607,
+       "partial_r_generated_given_broadcast": 0.14460913208606,
+       "r_generated_vs_broadcast": 0.3145193191198802,
+       "n_genes": 3277
+      },
+      "Monocyte": {
+       "r_generated": 0.18892332636666287,
+       "r_broadcast": 0.37567927431475395,
+       "partial_r_generated_given_broadcast": 0.058071878919912126,
+       "r_generated_vs_broadcast": 0.369783409371101,
+       "n_genes": 3152
+      },
+      "NK cell": {
+       "r_generated": 0.14299134222648124,
+       "r_broadcast": 0.2533010131298753,
+       "partial_r_generated_given_broadcast": 0.06018363378366271,
+       "r_generated_vs_broadcast": 0.3491259174809879,
+       "n_genes": 3069
+      }
+     },
+     "mean_r_generated": 0.18157664613193736,
+     "mean_r_broadcast": 0.263723888008903,
+     "mean_partial_r": 0.0972712819270106
+    },
+    "pseudobulk_log1p_cpm": {
+     "per_type": {
+      "T cell": {
+       "r_generated": 0.10685974295234307,
+       "r_broadcast": 0.29511657864157254,
+       "partial_r_generated_given_broadcast": 0.054508510436193425,
+       "r_generated_vs_broadcast": 0.1887916056253993,
+       "n_genes": 3072
+      },
+      "B cell": {
+       "r_generated": -0.014817159195365519,
+       "r_broadcast": 0.06708509712458424,
+       "partial_r_generated_given_broadcast": -0.023863605332527807,
+       "r_generated_vs_broadcast": 0.13099082120335118,
+       "n_genes": 3277
+      },
+      "Monocyte": {
+       "r_generated": 0.0882137599243738,
+       "r_broadcast": 0.33780407869585316,
+       "partial_r_generated_given_broadcast": 0.023774710288588918,
+       "r_generated_vs_broadcast": 0.19618312707273142,
+       "n_genes": 3152
+      },
+      "NK cell": {
+       "r_generated": 0.0959622271335853,
+       "r_broadcast": 0.1357496033139099,
+       "partial_r_generated_given_broadcast": 0.06976488289472567,
+       "r_generated_vs_broadcast": 0.20898288022567643,
+       "n_genes": 3069
+      }
+     },
+     "mean_r_generated": 0.06905464270373417,
+     "mean_r_broadcast": 0.20893883944397995,
+     "mean_partial_r": 0.031046124571745052
+    }
+   },
+   "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": 50,
+      "gen_top_list_size": 0,
+      "top50_overlap_broadcast": 4,
+      "top50_jaccard_broadcast": 0.05714285714285714,
+      "n_cells_obs": 42751,
+      "n_cells_gen": 21936
+     },
+     "B cell": {
+      "deg_scored": true,
+      "top50_overlap": 0,
+      "top50_jaccard": 0.0,
+      "obs_top_list_size": 50,
+      "gen_top_list_size": 0,
+      "top50_overlap_broadcast": 1,
+      "top50_jaccard_broadcast": 0.0136986301369863,
+      "n_cells_obs": 5672,
+      "n_cells_gen": 688
+     },
+     "Monocyte": {
+      "deg_scored": true,
+      "top50_overlap": 0,
+      "top50_jaccard": 0.0,
+      "obs_top_list_size": 50,
+      "gen_top_list_size": 0,
+      "top50_overlap_broadcast": 0,
+      "top50_jaccard_broadcast": 0.0,
+      "n_cells_obs": 28874,
+      "n_cells_gen": 8190
+     },
+     "NK cell": {
+      "deg_scored": true,
+      "top50_overlap": 0,
+      "top50_jaccard": 0.0,
+      "obs_top_list_size": 50,
+      "gen_top_list_size": 0,
+      "top50_overlap_broadcast": 0,
+      "top50_jaccard_broadcast": 0.0,
+      "n_cells_obs": 14784,
+      "n_cells_gen": 5511
+     },
+     "DC": {
+      "deg_scored": true,
+      "top50_overlap": 0,
+      "top50_jaccard": 0.0,
+      "obs_top_list_size": 50,
+      "gen_top_list_size": 0,
+      "top50_overlap_broadcast": 0,
+      "top50_jaccard_broadcast": 0.0,
+      "n_cells_obs": 1112,
+      "n_cells_gen": 1223
+     }
+    },
+    "sum": 0,
+    "sum_4type": 0,
+    "max": 250,
+    "mean": 0.0,
+    "sum_broadcast": 5,
+    "sum_broadcast_4type": 5,
+    "broadcast_n_significant": 24,
+    "seconds": 0.04
+   },
+   "seconds": 7.6
+  }
+ },
+ "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": 8.8
+}