Hugging Face's logo

Datasets:
junidude14
/
PBISC-data

Modalities:
Tabular
Text
Formats:
parquet
Languages:
English
Size:
< 1K
Tags:
single-cell-rna-seq
bulk-rna-seq
diffusion
Libraries:
Datasets
pandas
Polars
Dataset card Data Studio Files
xet
 Community
1
junidude14 commited on
Commit
4df3510
·
verified ·
1 Parent(s): 50878e1

Archive 4 checkpoints (1-4)

Browse files
Files changed (4) hide show
  1. paper-releases/pbisc-diffusion/updates/20261004T165054Z-rare-states-screen/diffusion/dataset/server_1/model8/plans/rare_states_screen_20261004/out/v1_provisional_markers/per_sample.csv +0 -0
  2. paper-releases/pbisc-diffusion/updates/20261004T165054Z-rare-states-screen/diffusion/dataset/server_1/model8/plans/rare_states_screen_20261004/out/v1_provisional_markers/screen_v1_stdout.txt +98 -0
  3. paper-releases/pbisc-diffusion/updates/20261004T165054Z-rare-states-screen/diffusion/dataset/server_1/model8/plans/rare_states_screen_20261004/out/v1_provisional_markers/summary.csv +37 -0
  4. paper-releases/pbisc-diffusion/updates/20261004T165054Z-rare-states-screen/diffusion/dataset/server_1/model8/plans/rare_states_screen_20261004/tools/corpus_states.py +77 -0
paper-releases/pbisc-diffusion/updates/20261004T165054Z-rare-states-screen/diffusion/dataset/server_1/model8/plans/rare_states_screen_20261004/out/v1_provisional_markers/per_sample.csv ADDED
The diff for this file is too large to render. See raw diff
 
paper-releases/pbisc-diffusion/updates/20261004T165054Z-rare-states-screen/diffusion/dataset/server_1/model8/plans/rare_states_screen_20261004/out/v1_provisional_markers/screen_v1_stdout.txt ADDED
@@ -0,0 +1,98 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ dataset state k thin gen obs_thinned obs_native_n gen_pct obs_pct gen_over_obs spearman
2
+ dress C1Q-high monocyte (reference) 2 0.607 110 / 3,793 529 / 4,485 586 2.900 11.795 0.246 0.964
3
+ dress Plasmablast (label) 1.000 76 / 35,000 187 / 33,712 187 0.217 0.555 0.391 0.571
4
+ dress Plasmablast (markers) 3 0.887 68 / 1,938 203 / 2,245 216 3.509 9.042 0.388 0.321
5
+ dress Progenitor (label) 1.000 20 / 35,000 35 / 33,712 35 0.057 0.104 0.550 0.432
6
+ dress Proliferating T (label) 1.000 1,001 / 27,578 4,226 / 24,482 4226 3.630 17.262 0.210 0.643
7
+ dress ABC / DN2 B 2 0.887 39 / 1,938 193 / 2,245 221 2.012 8.597 0.234 -0.214
8
+ dress GZMK+ CD8 T 1 1.000 1,138 / 6,901 838 / 6,038 838 16.490 13.879 1.188 0.964
9
+ dress Cytotoxic CD4 T 3 1.000 41 / 17,909 9 / 12,541 9 0.229 0.072 3.190 0.595
10
+ dress MS1 monocyte 3 0.607 184 / 3,793 417 / 4,485 797 4.851 9.298 0.522 0.679
11
+ dress ISG-high monocyte 4 0.607 12 / 3,793 131 / 4,485 242 0.316 2.921 0.108 0.517
12
+ dress Platelet-bound monocyte 2 0.607 6 / 3,793 26 / 4,485 41 0.158 0.580 0.273 0.576
13
+ dress Granulocyte-like (LDG) 2 0.967 0 / 35,000 404 / 33,712 421 0.000 1.198 0.000 NaN
14
+ kawasaki C1Q-high monocyte (reference) 2 1.000 43 / 7,927 603 / 11,050 603 0.542 5.457 0.099 0.173
15
+ kawasaki Plasmablast (label) 1.000 206 / 75,000 1,837 / 103,362 1837 0.275 1.777 0.155 0.558
16
+ kawasaki Plasmablast (markers) 3 1.000 237 / 20,998 1,654 / 21,517 1654 1.129 7.687 0.147 0.604
17
+ kawasaki Progenitor (label) 1.000 71 / 75,000 175 / 103,362 175 0.095 0.169 0.559 0.716
18
+ kawasaki Proliferating T (label) 1.000 330 / 42,408 1,146 / 57,032 1146 0.778 2.009 0.387 0.193
19
+ kawasaki ABC / DN2 B 2 1.000 397 / 20,998 907 / 21,517 907 1.891 4.215 0.449 -0.264
20
+ kawasaki GZMK+ CD8 T 1 1.000 1,884 / 11,370 746 / 15,371 746 16.570 4.853 3.414 0.343
21
+ kawasaki Cytotoxic CD4 T 3 1.000 94 / 26,586 163 / 28,856 163 0.354 0.565 0.626 0.461
22
+ kawasaki MS1 monocyte 3 1.000 1,442 / 7,927 3,263 / 11,050 3263 18.191 29.529 0.616 0.679
23
+ kawasaki ISG-high monocyte 4 1.000 280 / 7,927 2,256 / 11,050 2256 3.532 20.416 0.173 0.178
24
+ kawasaki Platelet-bound monocyte 2 1.000 27 / 7,927 649 / 11,050 649 0.341 5.873 0.058 0.367
25
+ kawasaki Granulocyte-like (LDG) 2 1.000 12 / 75,000 2,236 / 103,362 2236 0.016 2.163 0.007 0.794
26
+ sle C1Q-high monocyte (reference) 2 0.957 109 / 21,833 1,798 / 70,901 1823 0.499 2.536 0.197 0.404
27
+ sle Plasmablast (label) 1.000 85 / 168,000 1,526 / 358,235 1526 0.051 0.426 0.119 0.190
28
+ sle Plasmablast (markers) 3 1.000 72 / 19,311 1,298 / 35,488 1298 0.373 3.658 0.102 0.148
29
+ sle Progenitor (label) 1.000 35 / 168,000 352 / 358,235 352 0.021 0.098 0.212 0.335
30
+ sle Proliferating T (label) 1.000 561 / 106,521 1,953 / 216,491 1953 0.527 0.902 0.584 0.670
31
+ sle ABC / DN2 B 2 1.000 103 / 19,311 2,213 / 35,488 2213 0.533 6.236 0.086 -0.272
32
+ sle GZMK+ CD8 T 1 1.000 6,044 / 22,345 9,083 / 69,054 9083 27.049 13.153 2.056 0.450
33
+ sle Cytotoxic CD4 T 3 1.000 432 / 71,556 1,471 / 106,194 1471 0.604 1.385 0.436 0.639
34
+ sle MS1 monocyte 3 0.957 381 / 21,833 8,977 / 70,901 9480 1.745 12.661 0.138 0.044
35
+ sle ISG-high monocyte 4 0.957 4,943 / 21,833 20,677 / 70,901 21269 22.640 29.163 0.776 0.837
36
+ sle Platelet-bound monocyte 2 0.957 209 / 21,833 2,259 / 70,901 2315 0.957 3.186 0.300 0.150
37
+ sle Granulocyte-like (LDG) 2 1.000 9 / 168,000 2,107 / 358,235 2107 0.005 0.588 0.009 0.188
38
+ dataset state contrast obs_a_counts obs_b_counts gen_a_counts gen_b_counts obs_diff gen_diff same_direction
39
+ dress C1Q-high monocyte (reference) DRESS|none - healthy|none 53 / 1388 32 / 1949 15 / 705 11 / 1148 2.18 1.17 True
40
+ dress Plasmablast (label) DRESS|none - healthy|none 43 / 7592 2 / 8588 12 / 5000 2 / 5000 0.54 0.20 True
41
+ dress Plasmablast (markers) DRESS|none - healthy|none 48 / 375 12 / 663 15 / 326 2 / 327 10.99 3.99 True
42
+ dress Progenitor (label) DRESS|none - healthy|none 6 / 7592 11 / 8588 7 / 5000 3 / 5000 -0.05 0.08 False
43
+ dress Proliferating T (label) DRESS|none - healthy|none 2361 / 5394 59 / 4831 70 / 3696 5 / 2875 42.55 1.72 True
44
+ dress ABC / DN2 B DRESS|none - healthy|none 21 / 375 45 / 663 12 / 326 5 / 327 -1.19 2.15 False
45
+ dress GZMK+ CD8 T DRESS|none - healthy|none 17 / 696 764 / 2085 233 / 894 551 / 1496 -34.20 -10.77 True
46
+ dress Cytotoxic CD4 T DRESS|none - healthy|none 1 / 2108 2 / 2003 12 / 2477 7 / 907 -0.05 -0.29 True
47
+ dress MS1 monocyte DRESS|none - healthy|none 166 / 1388 129 / 1949 79 / 705 47 / 1148 5.34 7.11 True
48
+ dress ISG-high monocyte DRESS|none - healthy|none 26 / 1388 49 / 1949 6 / 705 3 / 1148 -0.64 0.59 False
49
+ dress Platelet-bound monocyte DRESS|none - healthy|none 12 / 1388 13 / 1949 2 / 705 2 / 1148 0.20 0.11 True
50
+ dress Granulocyte-like (LDG) DRESS|none - healthy|none 222 / 7592 152 / 8588 0 / 5000 0 / 5000 1.15 0.00 False
51
+ kawasaki C1Q-high monocyte (reference) KD_before - HC 344 / 5923 45 / 1123 21 / 4450 1 / 798 2.92 0.35 True
52
+ kawasaki C1Q-high monocyte (reference) KD_before - KD_after 344 / 5923 214 / 4004 21 / 4450 21 / 2679 3.05 -0.36 False
53
+ kawasaki Plasmablast (label) KD_before - HC 498 / 36717 116 / 16203 76 / 30000 24 / 15000 0.73 0.09 True
54
+ kawasaki Plasmablast (label) KD_before - KD_after 498 / 36717 1223 / 50442 76 / 30000 106 / 30000 -1.00 -0.10 True
55
+ kawasaki Plasmablast (markers) KD_before - HC 576 / 11634 156 / 3316 93 / 11002 28 / 4266 0.87 0.35 True
56
+ kawasaki Plasmablast (markers) KD_before - KD_after 576 / 11634 922 / 6567 93 / 11002 116 / 5730 -8.73 -1.06 True
57
+ kawasaki Progenitor (label) KD_before - HC 58 / 36717 7 / 16203 46 / 30000 0 / 15000 0.12 0.15 True
58
+ kawasaki Progenitor (label) KD_before - KD_after 58 / 36717 110 / 50442 46 / 30000 25 / 30000 -0.03 0.07 False
59
+ kawasaki Proliferating T (label) KD_before - HC 158 / 16366 164 / 10135 125 / 13373 50 / 9110 -0.59 0.37 False
60
+ kawasaki Proliferating T (label) KD_before - KD_after 158 / 16366 824 / 30531 125 / 13373 155 / 19925 -1.80 0.13 False
61
+ kawasaki ABC / DN2 B KD_before - HC 457 / 11634 167 / 3316 213 / 11002 68 / 4266 -0.59 0.30 False
62
+ kawasaki ABC / DN2 B KD_before - KD_after 457 / 11634 283 / 6567 213 / 11002 116 / 5730 -0.27 -0.22 True
63
+ kawasaki GZMK+ CD8 T KD_before - HC 152 / 4930 238 / 3652 488 / 3518 583 / 3062 -3.68 -4.41 True
64
+ kawasaki GZMK+ CD8 T KD_before - KD_after 152 / 4930 356 / 6789 488 / 3518 813 / 4790 -3.05 -2.63 True
65
+ kawasaki Cytotoxic CD4 T KD_before - HC 5 / 8948 18 / 4852 27 / 8516 17 / 4764 -0.30 -0.02 True
66
+ kawasaki Cytotoxic CD4 T KD_before - KD_after 5 / 8948 140 / 15056 27 / 8516 50 / 13306 -0.72 -0.08 True
67
+ kawasaki MS1 monocyte KD_before - HC 2642 / 5923 50 / 1123 1087 / 4450 39 / 798 30.74 11.81 True
68
+ kawasaki MS1 monocyte KD_before - KD_after 2642 / 5923 571 / 4004 1087 / 4450 316 / 2679 20.63 5.64 True
69
+ kawasaki ISG-high monocyte KD_before - HC 395 / 5923 254 / 1123 31 / 4450 3 / 798 -15.02 0.11 False
70
+ kawasaki ISG-high monocyte KD_before - KD_after 395 / 5923 1607 / 4004 31 / 4450 246 / 2679 -12.20 -4.79 True
71
+ kawasaki Platelet-bound monocyte KD_before - HC 321 / 5923 65 / 1123 14 / 4450 2 / 798 -2.81 0.07 False
72
+ kawasaki Platelet-bound monocyte KD_before - KD_after 321 / 5923 263 / 4004 14 / 4450 11 / 2679 -1.82 -0.13 True
73
+ kawasaki Granulocyte-like (LDG) KD_before - HC 1083 / 36717 45 / 16203 8 / 30000 0 / 15000 3.36 0.03 True
74
+ kawasaki Granulocyte-like (LDG) KD_before - KD_after 1083 / 36717 1108 / 50442 8 / 30000 4 / 30000 1.28 0.01 True
75
+ sle C1Q-high monocyte (reference) aSLE - aHD 348 / 9639 108 / 6619 21 / 3390 7 / 1665 1.62 -0.10 False
76
+ sle C1Q-high monocyte (reference) cSLE - cHD 1225 / 45551 117 / 9092 77 / 14947 4 / 1831 0.58 0.11 True
77
+ sle Plasmablast (label) aSLE - aHD 143 / 35392 77 / 41440 3 / 21000 6 / 15000 0.18 -0.03 False
78
+ sle Plasmablast (label) cSLE - cHD 1084 / 195838 222 / 85565 69 / 99000 7 / 33000 0.39 0.05 True
79
+ sle Plasmablast (markers) aSLE - aHD 134 / 936 68 / 2442 3 / 911 5 / 1221 13.65 -0.16 False
80
+ sle Plasmablast (markers) cSLE - cHD 907 / 21628 189 / 10482 57 / 12436 7 / 4743 2.95 0.27 True
81
+ sle Progenitor (label) aSLE - aHD 50 / 35392 60 / 41440 6 / 21000 0 / 15000 -0.01 0.03 False
82
+ sle Progenitor (label) cSLE - cHD 175 / 195838 67 / 85565 26 / 99000 3 / 33000 0.01 0.02 True
83
+ sle Proliferating T (label) aSLE - aHD 235 / 20730 69 / 27704 108 / 12979 13 / 10265 1.12 0.85 True
84
+ sle Proliferating T (label) cSLE - cHD 1532 / 111977 117 / 56080 426 / 60156 14 / 23121 1.84 0.73 True
85
+ sle ABC / DN2 B aSLE - aHD 65 / 936 85 / 2442 3 / 911 8 / 1221 4.71 -0.38 False
86
+ sle ABC / DN2 B cSLE - cHD 1669 / 21628 394 / 10482 63 / 12436 29 / 4743 4.71 -0.14 False
87
+ sle GZMK+ CD8 T aSLE - aHD 1559 / 7125 1207 / 5039 966 / 2969 609 / 1747 -0.75 -1.97 True
88
+ sle GZMK+ CD8 T cSLE - cHD 5363 / 39111 954 / 17779 3448 / 13518 1021 / 4111 7.49 0.47 True
89
+ sle Cytotoxic CD4 T aSLE - aHD 144 / 9990 96 / 17255 75 / 7915 43 / 7211 2.28 0.59 True
90
+ sle Cytotoxic CD4 T cSLE - cHD 869 / 51287 362 / 27662 230 / 39554 84 / 16876 1.31 0.10 True
91
+ sle MS1 monocyte aSLE - aHD 644 / 9639 648 / 6619 70 / 3390 17 / 1665 -1.35 0.76 False
92
+ sle MS1 monocyte cSLE - cHD 7017 / 45551 668 / 9092 262 / 14947 32 / 1831 6.54 -0.50 False
93
+ sle ISG-high monocyte aSLE - aHD 3051 / 9639 105 / 6619 778 / 3390 12 / 1665 31.00 18.82 True
94
+ sle ISG-high monocyte cSLE - cHD 17241 / 45551 280 / 9092 4134 / 14947 19 / 1831 32.39 22.75 True
95
+ sle Platelet-bound monocyte aSLE - aHD 386 / 9639 95 / 6619 35 / 3390 14 / 1665 2.54 0.53 True
96
+ sle Platelet-bound monocyte cSLE - cHD 1400 / 45551 378 / 9092 150 / 14947 10 / 1831 -0.98 0.41 False
97
+ sle Granulocyte-like (LDG) aSLE - aHD 216 / 35392 136 / 41440 2 / 21000 0 / 15000 0.29 0.01 True
98
+ sle Granulocyte-like (LDG) cSLE - cHD 1531 / 195838 224 / 85565 6 / 99000 1 / 33000 0.47 0.00 True
paper-releases/pbisc-diffusion/updates/20261004T165054Z-rare-states-screen/diffusion/dataset/server_1/model8/plans/rare_states_screen_20261004/out/v1_provisional_markers/summary.csv ADDED
@@ -0,0 +1,37 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ dataset,state,genes_used,k,thin,gen,obs_thinned,obs_native_n,gen_pct,obs_pct,gen_over_obs,spearman
2
+ dress,C1Q-high monocyte (reference),C1QA C1QB C1QC,2,0.607,"110 / 3,793","529 / 4,485",586,2.9000790930661746,11.794871794871796,0.24587627093387132,0.9642857142857145
3
+ dress,Plasmablast (label),,,1.0,"76 / 35,000","187 / 33,712",187,0.21714285714285714,0.5546986236355007,0.39146096256684493,0.5714285714285715
4
+ dress,Plasmablast (markers),MZB1 JCHAIN TNFRSF17 PRDM1 TXNDC5,3,0.887,"68 / 1,938","203 / 2,245",216,3.508771929824561,9.042316258351892,0.38803906317517933,0.3214285714285715
5
+ dress,Progenitor (label),,,1.0,"20 / 35,000","35 / 33,712",35,0.05714285714285714,0.10382059800664452,0.5504,0.43244998209386837
6
+ dress,Proliferating T (label),,,1.0,"1,001 / 27,578","4,226 / 24,482",4226,3.6297048371890637,17.261661628951884,0.2102755178042183,0.642857142857143
7
+ dress,ABC / DN2 B,ITGAX TBX21 FCRL5 ZEB2,2,0.887,"39 / 1,938","193 / 2,245",221,2.0123839009287927,8.596881959910913,0.23408299780233885,-0.21428571428571433
8
+ dress,GZMK+ CD8 T,GZMK,1,1.0,"1,138 / 6,901","838 / 6,038",838,16.490363715403564,13.878767803908579,1.188172029995307,0.9642857142857145
9
+ dress,Cytotoxic CD4 T,GZMB PRF1 GNLY GZMH NKG7,3,1.0,"41 / 17,909","9 / 12,541",9,0.22893517225975765,0.07176461207240252,3.190084439232912,0.594618725379069
10
+ dress,MS1 monocyte,RETN ALOX5AP IL1R2 CYP1B1 MCEMP1 PLAC8,3,0.607,"184 / 3,793","417 / 4,485",797,4.851041392037964,9.297658862876254,0.521748696481781,0.6785714285714287
11
+ dress,ISG-high monocyte,IFI27 IFI44L ISG15 MX1 RSAD2 SIGLEC1 IFIT1 IFIT3 OAS1,4,0.607,"12 / 3,793","131 / 4,485",242,0.3163722646981281,2.9208472686733558,0.10831523718863394,0.5169078021169037
12
+ dress,Platelet-bound monocyte,PPBP PF4 TUBB1 GP9,2,0.607,"6 / 3,793","26 / 4,485",41,0.15818613234906406,0.5797101449275363,0.2728710783021355,0.576086956521739
13
+ dress,Granulocyte-like (LDG),ELANE MPO CAMP LCN2 CSF3R CXCR2,2,0.967,"0 / 35,000","404 / 33,712",421,0.0,1.1983863312766967,0.0,
14
+ kawasaki,C1Q-high monocyte (reference),C1QA C1QB C1QC,2,1.0,"43 / 7,927","603 / 11,050",603,0.5424498549262016,5.457013574660634,0.09940416081151786,0.17297999283754734
15
+ kawasaki,Plasmablast (label),,,1.0,"206 / 75,000","1,837 / 103,362",1837,0.27466666666666667,1.7772488922427971,0.15454597713663581,0.5583563931615944
16
+ kawasaki,Plasmablast (markers),MZB1 JCHAIN TNFRSF17 PRDM1 TXNDC5,3,1.0,"237 / 20,998","1,654 / 21,517",1654,1.1286789218020763,7.6869452061160946,0.1468306188658723,0.6035714285714285
17
+ kawasaki,Progenitor (label),,,1.0,"71 / 75,000","175 / 103,362",175,0.09466666666666666,0.1693078694297711,0.5591392,0.7164951568756692
18
+ kawasaki,Proliferating T (label),,,1.0,"330 / 42,408","1,146 / 57,032",1146,0.77815506508206,2.009398232571188,0.38725776327888345,0.1928571428571428
19
+ kawasaki,ABC / DN2 B,ITGAX TBX21 FCRL5 ZEB2,2,1.0,"397 / 20,998","907 / 21,517",907,1.8906562529764739,4.215271645675513,0.44852536488748385,-0.2642857142857143
20
+ kawasaki,GZMK+ CD8 T,GZMK,1,1.0,"1,884 / 11,370","746 / 15,371",746,16.569920844327175,4.853295166222106,3.414158891396153,0.3428571428571428
21
+ kawasaki,Cytotoxic CD4 T,GZMB PRF1 GNLY GZMH NKG7,3,1.0,"94 / 26,586","163 / 28,856",163,0.35356954788234407,0.5648738563903521,0.6259265566682772,0.4612799809001262
22
+ kawasaki,MS1 monocyte,RETN ALOX5AP IL1R2 CYP1B1 MCEMP1 PLAC8,3,1.0,"1,442 / 7,927","3,263 / 11,050",3263,18.190992809385644,29.529411764705884,0.6160296369712269,0.6785714285714285
23
+ kawasaki,ISG-high monocyte,IFI27 IFI44L ISG15 MX1 RSAD2 SIGLEC1 IFIT1 IFIT3 OAS1,4,1.0,"280 / 7,927","2,256 / 11,050",2256,3.532231613472941,20.41628959276018,0.17301045801806736,0.17821127702606046
24
+ kawasaki,Platelet-bound monocyte,PPBP PF4 TUBB1 GP9,2,1.0,"27 / 7,927","649 / 11,050",649,0.3406080484420336,5.873303167420814,0.05799258760068523,0.3673334485639205
25
+ kawasaki,Granulocyte-like (LDG),ELANE MPO CAMP LCN2 CSF3R CXCR2,2,1.0,"12 / 75,000","2,236 / 103,362",2236,0.016,2.163270834542675,0.007396207513416816,0.7940844788408192
26
+ sle,C1Q-high monocyte (reference),C1QA C1QB C1QC,2,0.957,"109 / 21,833","1,798 / 70,901",1823,0.499244263271195,2.5359303817999748,0.19686828426135147,0.40405326166118033
27
+ sle,Plasmablast (label),,,1.0,"85 / 168,000","1,526 / 358,235",1526,0.050595238095238096,0.42597736122936064,0.11877447653373276,0.19002519863132827
28
+ sle,Plasmablast (markers),MZB1 TNFRSF17 PRDM1 TXNDC5,3,1.0,"72 / 19,311","1,298 / 35,488",1298,0.37284449277613796,3.6575743913435526,0.10193763759352531,0.14813086865365177
29
+ sle,Progenitor (label),,,1.0,"35 / 168,000","352 / 358,235",352,0.020833333333333332,0.09825952238056025,0.2120235558712121,0.33472476145258456
30
+ sle,Proliferating T (label),,,1.0,"561 / 106,521","1,953 / 216,491",1953,0.5266567155772102,0.9021160232988901,0.5838015310395587,0.670138401025515
31
+ sle,ABC / DN2 B,ITGAX TBX21 FCRL5 ZEB2,2,1.0,"103 / 19,311","2,213 / 35,488",2213,0.5333747604991973,6.235910730387737,0.08553277677630146,-0.2718355039783739
32
+ sle,GZMK+ CD8 T,GZMK,1,1.0,"6,044 / 22,345","9,083 / 69,054",9083,27.048556724099353,13.153474092739016,2.056381191264953,0.44976076555023925
33
+ sle,Cytotoxic CD4 T,GZMB PRF1 GNLY GZMH NKG7,3,1.0,"432 / 71,556","1,471 / 106,194",1471,0.6037229582424953,1.3852006704710247,0.43583790501430014,0.6385618347110514
34
+ sle,MS1 monocyte,RETN ALOX5AP IL1R2 CYP1B1 PLAC8,3,0.957,"381 / 21,833","8,977 / 70,901",9480,1.7450648101497732,12.661316483547482,0.13782649003501066,0.043753953156657185
35
+ sle,ISG-high monocyte,IFI27 IFI44L ISG15 MX1 RSAD2 SIGLEC1 IFIT1 IFIT3 OAS1,4,0.957,"4,943 / 21,833","20,677 / 70,901",21269,22.64004030595887,29.16319939069971,0.7763222410082651,0.8368137881033272
36
+ sle,Platelet-bound monocyte,PPBP PF4 TUBB1 GP9,2,0.957,"209 / 21,833","2,259 / 70,901",2315,0.9572665231530252,3.186132776688622,0.30044778113356635,0.1503215921733296
37
+ sle,Granulocyte-like (LDG),ELANE MPO CAMP LCN2 CSF3R CXCR2,2,1.0,"9 / 168,000","2,107 / 358,235",2107,0.005357142857142857,0.5881614024313648,0.009108287002508646,0.18801031804872972
paper-releases/pbisc-diffusion/updates/20261004T165054Z-rare-states-screen/diffusion/dataset/server_1/model8/plans/rare_states_screen_20261004/tools/corpus_states.py ADDED
@@ -0,0 +1,77 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env python3
2
+ """How common is each rare state of tools/states.json in the pretraining corpus? (CPU only; 2026-10-04)
3
+
4
+ Re-uses the corpus sample of plans/behcet_c1q_lora3_20261004 (corpus_c1q.py extract: 179,750 cells, random blocks of
5
+ every one of the 22 corpus datasets, 6k panel; frozen f01common scANVI labels in work/corpus_c1q/scanvi_labels). The
6
+ h5ad itself lives on spark-0110 (sha-checked .moved marker); pass a local copy. Rates are at native depth (no
7
+ thinning) and use the same state rules as screen.py. A gene is counted only in datasets where it was measured
8
+ (column sum > 0 in that dataset's sample); a state with a missing gene in a dataset is flagged.
9
+ usage: corpus_states.py CORPUS_SAMPLE_H5AD output: out/corpus_states.csv
10
+ """
11
+ from __future__ import annotations
12
+
13
+ import json
14
+ import sys
15
+ from pathlib import Path
16
+
17
+ import anndata as ad
18
+ import numpy as np
19
+ import pandas as pd
20
+ import scipy.sparse as sp
21
+
22
+ sys.path.insert(0, str(Path(__file__).resolve().parent))
23
+ from screen import call_state, state_cols # noqa: E402 same state rules as the generated-vs-observed screen
24
+
25
+ PLAN = Path(__file__).resolve().parents[1]
26
+ LABELS = PLAN.parent / "behcet_c1q_lora3_20261004/work/corpus_c1q/scanvi_labels/labels.parquet"
27
+ CANON = Path("/home/sj_server_1/gblee/datalake/paired-pbmc-bulk-sc/processing/strict_qc_v1_20260913/reference/"
28
+ "canonical_genes.tsv")
29
+
30
+
31
+ from screen import parent_mask # noqa: E402
32
+
33
+
34
+ def main() -> int:
35
+ a = ad.read_h5ad(sys.argv[1])
36
+ lab = pd.read_parquet(LABELS)
37
+ if len(lab) != a.n_obs:
38
+ raise SystemExit("labels / cells differ")
39
+ states = json.loads((PLAN / "tools/states.json").read_text())
40
+ canon = pd.read_csv(CANON, sep="\t", dtype=str)
41
+ sym2id = dict(zip(canon.gene_symbol, canon.gene_id))
42
+ pos = {g: i for i, g in enumerate(a.var_names.astype(str))}
43
+ X = sp.csr_matrix(a.X)
44
+ ds = a.obs.dataset_id.astype(str).to_numpy()
45
+ rows = []
46
+ for d in pd.unique(ds):
47
+ m = ds == d
48
+ Xd = X[m]
49
+ measured = np.asarray(Xd.sum(0)).ravel() > 0
50
+ ld = lab[m].reset_index(drop=True)
51
+ for st in states:
52
+ pm = parent_mask(ld, st["parent"])
53
+ rec = {"dataset_id": d, "state": st["name"], "parent_cells": int(pm.sum())}
54
+ if st["kind"] == "label":
55
+ h = ld.aifi_l2.isin(st["labels"]).to_numpy() & pm
56
+ rec["missing_genes"] = ""
57
+ else:
58
+ genes, pcols, req, exc = state_cols(st, sym2id, pos)
59
+ rec["missing_genes"] = " ".join(g for g, c in zip(genes, pcols) if not measured[c])
60
+ cols = pcols + [c for r in req for c in r] + exc
61
+ h = call_state(Xd[:, cols].toarray(), st, len(pcols), req, exc) & pm
62
+ rec["state_cells"] = int(h.sum())
63
+ rec["pct"] = 100 * rec["state_cells"] / rec["parent_cells"] if rec["parent_cells"] else np.nan
64
+ rows.append(rec)
65
+ t = pd.DataFrame(rows)
66
+ t.to_csv(PLAN / "out/corpus_states.csv", index=False)
67
+ w = t.assign(short=t.dataset_id.str.slice(0, 40)).pivot(index="short", columns="state", values="pct")
68
+ tot = t.groupby("state")[["state_cells", "parent_cells"]].sum()
69
+ with pd.option_context("display.width", 300, "display.max_columns", 30):
70
+ print(w.round(2).to_string())
71
+ print((tot.assign(pct=100 * tot.state_cells / tot.parent_cells)).round(3).to_string())
72
+ print(t[t.missing_genes != ""][["dataset_id", "state", "missing_genes"]].to_string(index=False))
73
+ return 0
74
+
75
+
76
+ if __name__ == "__main__":
77
+ sys.exit(main())