File size: 39,370 Bytes
a48f0ae
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
# ---
# jupyter:
#   jupytext:
#     text_representation:
#       extension: .py
#       format_name: light
#       format_version: '1.5'
#       jupytext_version: 1.16.1
#   kernelspec:
#     display_name: chemical_CPA
#     language: python
#     name: python3
# ---

# # 5 SCIPLEX OOD SPLITS
#
# **Requires**
# * `'trapnell_cpa_lincs_genes.h5ad'`
# * `'trapnell_cpa.h5ad'`
#
# **Output**
# * `'sciplex_complete.h5ad'`
# * `'sciplex_complete_lincs_genes.h5ad'`
#
#
# ## Description
#
# The main purpose of this notebook is to create various subdatasets and splits of the data created in previous notebooks. 
#
# ### Datasets
#
# The complete set of datasets created throughout this notebook is listed in the following table:
#
# | **Dataset Name**                                         | **Purpose**                                      | **Includes Splits?**                                                | **Notes**                                                                          |
# | -------------------------------------------------------- | ------------------------------------------------ | ------------------------------------------------------------------- | ---------------------------------------------------------------------------------- |
# | **`sciplex_complete.h5ad`**                              | Complete dataset                                 | No                                                                  | Full dataset without any modifications.                                            |
# | **`sciplex_complete_lincs_genes.h5ad`**                  | Complete dataset focusing on LINCS genes         | No                                                                  | Similar to above but specific to LINCS genes.                                      |
# | **`adata_MCF7.h5ad`**                                    | Cell-type specific (MCF7) dataset                | No                                                                  | Subset of `sciplex_complete.h5ad` for MCF7 cells.                                  |
# | **`adata_MCF7_lincs_genes.h5ad`**                        | LINCS-specific MCF7 dataset                      | No                                                                  | Subset of `sciplex_complete_lincs_genes.h5ad` for MCF7 cells.                      |
# | **`adata_K562.h5ad`**                                    | Cell-type specific (K562) dataset                | No                                                                  | Subset of `sciplex_complete.h5ad` for K562 cells.                                  |
# | **`adata_K562_lincs_genes.h5ad`**                        | LINCS-specific K562 dataset                      | No                                                                  | Subset of `sciplex_complete_lincs_genes.h5ad` for K562 cells.                      |
# | **`adata_A549.h5ad`**                                    | Cell-type specific (A549) dataset                | No                                                                  | Subset of `sciplex_complete.h5ad` for A549 cells.                                  |
# | **`adata_A549_lincs_genes.h5ad`**                        | LINCS-specific A549 dataset                      | No                                                                  | Subset of `sciplex_complete_lincs_genes.h5ad` for A549 cells.                      |
# | **`sciplex_complete_subset_lincs_genes_v2.h5ad`**        | Small dataset with 40 observations per condition | Partially (`split`)                                                 | Randomly subsampled, basic splits (`train`, `test`, `ood`) added.                  |
# | **`sciplex_complete_middle_subset_v2.h5ad`**             | Middle-sized dataset with dosage diversity       | Partially (`split_ood_finetuning`)                                  | Observations subsampled by dosage, basic splits (`ood`, `train`, `test`) included. |
# | **`sciplex_complete_middle_subset_lincs_genes_v2.h5ad`** | Middle-sized LINCS-specific dataset              | Partially (`split_ood_finetuning`)                                  | Same as above but focused on LINCS genes.                                          |
# | **`sciplex_complete_v2.h5ad`**                           | Final processed dataset                          | Yes (`split_ood_finetuning`, `split_ho_epigenetic`, `split_random`) | Fully annotated with all splits.                                                   |
# | **`sciplex_complete_lincs_genes_v2.h5ad`**               | Final LINCS-specific processed dataset           | Yes (`split_ood_finetuning`, `split_ho_epigenetic`, `split_random`) | Fully annotated with all splits and LINCS gene focus.                              |
#
#
#
# ### Splits
#
# For convenience, we also describe how the splits are created, but you may also look into the code below
#
# ### `split_ood_finetuning` (Main Train/Test/OOD Split)
#
# This split simply divides the entire dataset into **train**, **test**, and **OOD** categories, as follows:
#
# - **Train**: All rows are initially assigned to `'train'`.
#
# - **OOD**: Rows where the drug (`condition`) is one of the predefined **OOD drugs** (`Dacinostat`, `CUDC-907`, `Givinostat`, `CUDC-101`, `Pirarubicin`, `Hesperadin`, `Tanespimycin`, `Trametinib`, `Raltitrexed`) are reassigned to `'ood'`.
#
# - **Test**: Rows are assigned to `'test'` in three steps:
#
#   1. **Validation Drugs**: 40% of rows for specific validation drugs at high doses (`1e3` or `1e4`) and 20% for lower doses (`1e1` or `1e2`).
#   2. **Random Sample**: 4% of remaining `'train'` rows are moved to `'test'`.
#   3. **Control Rows**: 5% of control rows (`control == 1`) are moved to `'test'`.
#
# ### 2. `split_ho_epigenetic` (Epigenetic Holdout Split)
#
# The validation and OOD splits in this dataset focus on epigenetic regulation. They are sampled by using a predefined selection of epigenetic drugs:
#
# - **Validation Epigenetic Drugs**
#
#   - Trichostatin, CUDC-101, M344, Resminostat, Entinostat, Tucidinostat, Tacedinaline, Mocetinostat, Pracinostat.
#
# - **Out-of-Distribution Epigenetic Drugs**
#
#   - Dacinostat, Quisinostat, CUDC-907, Abexinostat, Panobinostat, Belinostat, Givinostat, AR-42.
#
# These splits are then defined as follows:
#
# - **Train**: All rows initially assigned to `'train'`
# - **Test**: Rows are moved to `'test'` in several steps:
#   1. **Validation Drugs for High Doses**: 40% of the samples with validation epigenetic drugs at high doses (`1e3` or `1e4`) are moved to `'test'`.
#   2. **Validation Drugs for Low Doses**: 40% of the samples with validation epigenetic drugs at lower doses (`1e1` or `1e2`) are also moved to `'test'`.
#   3. **Remaining Epigenetic Drugs**: 40% of samples for epigenetic drugs that are not part of the primary training set (`epigenetic_drugs_all` excluding `epigenetic_drugs`) are moved to `'test'`.
#   4. **Control Samples**: 40% of the control samples (`control == 1`) are moved to `'test'` to ensure proper evaluation of untreated conditions.
# - **OOD**: Observations treated with any drug inside `ood_epi_drugs` are reassigned to `'ood'`.
#
# ### 3. `split_ho_epigenetic_all` (All Epigenetic Holdout Split)
#
# Similar to `split_ho_epigenetic`, but this split users the **remaining epigenetic drugs that are not manually selected** in section 2. Specifically:
#
# The set of drugs used in this split (`epigenetic_drugs_all`) is defined as all unique drugs within the dataset that are associated with the pathway **'Epigenetic regulation'**. This is determined by selecting all drugs for which the pathway information (`pathway_level_1`) is labeled as **'Epigenetic regulation'**.
#
# - **Train**: Initially, all rows are assigned to `'train'`.
#
# - **Test**: Rows are reassigned to `'test'` in several steps:
#
#   1. **Initial Split**: 5% of the entire dataset is randomly selected and reassigned to `'test'`.
#   2. **Epigenetic Drug Validation**: 50% of samples treated with epigenetic drugs that are not in the main training set (`epigenetic_drugs_all` excluding `epigenetic_drugs`) are reassigned to `'test'`.
#   3. **Control Samples**: 40% of control samples (`control == 1`) are reassigned to `'test'` to ensure proper evaluation of untreated conditions.
#
# - **OOD**: Observations treated with any drug within `epigenetic_drugs` are reassigned to `'ood'`.
#
# ### 4. `split_random` (Random Split)
#
# The entire dataset is divided as follows:
#
# - **Train**: 70% of the rows are randomly assigned to `'train'`.
# - **Test**: 15% of the rows are randomly assigned to `'test'`.
# - **OOD**: 15% of the rows are randomly assigned to `'ood'`. This split serves as a straightforward baseline, ensuring a random distribution of conditions across the splits without specific considerations for pathways or drug types.
#
#

# ## Imports

# +
import os
import logging

import matplotlib
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import scanpy as sc
import seaborn as sns
import sfaira
from rdkit import Chem, DataStructs
from rdkit.Chem import Draw
from rdkit.Chem.Draw import IPythonConsole

from chemCPA.paths import DATA_DIR, PROJECT_DIR

IPythonConsole.ipython_useSVG = False
matplotlib.style.use("fivethirtyeight")
# matplotlib.style.use("seaborn-talk")
matplotlib.rcParams['font.family'] = "monospace"
matplotlib.rcParams['figure.dpi'] = 200
matplotlib.pyplot.rcParams['savefig.facecolor'] = 'white'
os.getcwd()
pd.set_option('display.max_columns', 100)
sc.set_figure_params(dpi=80, frameon=False)
sc.logging.print_header()
sns.set_context("poster")
import rapids_singlecell as rsc

logging.basicConfig(level=logging.INFO)

# -

# %load_ext autoreload
# %autoreload 2

# ## Load data

logging.info("Starting to load in data from %s", PROJECT_DIR/'datasets'/'trapnell_cpa_lincs_genes.h5ad')
adata_sciplex_lincs_genes = sc.read(PROJECT_DIR/'datasets'/'trapnell_cpa_lincs_genes.h5ad')
logging.info("Data loaded from %s", PROJECT_DIR/'datasets'/'trapnell_cpa_lincs_genes.h5ad')
logging.info("Starting to load in data from %s", PROJECT_DIR/'datasets'/'trapnell_cpa.h5ad')
adata_sciplex = sc.read(PROJECT_DIR/'datasets'/'trapnell_cpa.h5ad')
logging.info("Data loaded from %s", PROJECT_DIR/'datasets'/'trapnell_cpa.h5ad')

# +
if 'log1p' in adata_sciplex.uns:
    del adata_sciplex.uns['log1p']

if 'log1p' in adata_sciplex_lincs_genes.uns:
    del adata_sciplex_lincs_genes.uns['log1p']
# -

# # Compute highly variable genes

# +
import cupy as cp

# Convert to dense numpy array and then to cupy array
logging.info("Starting computation of highly variable genes for adata_sciplex_lincs_genes")
adata_sciplex_lincs_genes.X = cp.array(adata_sciplex_lincs_genes.X.toarray())
rsc.pp.highly_variable_genes(adata_sciplex_lincs_genes, n_top_genes=977)
logging.info("Finished computation of highly variable genes for adata_sciplex_lincs_genes")
# -

logging.info("Starting computation of highly variable genes for adata_sciplex")
adata_sciplex.X = cp.array(adata_sciplex.X.toarray())
rsc.pp.highly_variable_genes(adata_sciplex, n_top_genes=2000)
logging.info("Finished computation of highly variable genes for adata_sciplex")


# ## Compute UMAP

# ! mamba list | grep pynndescent
# ! mamba list | grep umap

# +
def preprocess_adata(adata, n_comps=25, n_neighbors=50):
    rsc.pp.pca(adata, n_comps=n_comps)
    rsc.pp.neighbors(adata, n_neighbors=n_neighbors, metric='cosine')
    rsc.tl.umap(adata, min_dist=0.1)
    return None

def preprocess_adata_subset_type(adata, cell_type, n_comps=25):
    adata_new = adata[adata.obs.cell_type == cell_type].copy()
    rsc.pp.pca(adata_new, n_comps=n_comps)
    rsc.pp.neighbors(adata_new, n_neighbors=50, metric='cosine')
    rsc.tl.umap(adata_new, min_dist=0.1)
    return adata_new


# -

# This can take >20min to process
logging.info("Starting preprocessing of adata_sciplex")
preprocess_adata(adata_sciplex, n_comps=25, n_neighbors=50)
logging.info("Completed preprocessing of adata_sciplex")

# This can take >20min to process
logging.info("Starting preprocessing of adata_sciplex_lincs_genes")
preprocess_adata(adata_sciplex_lincs_genes)
logging.info("Completed preprocessing of adata_sciplex_lincs_genes")

# ## Load or create subsetted adata objects

# +
# # +
logging.info("Starting to load/create subsetted adata objects")

# MCF7
fname = PROJECT_DIR/'datasets'/'adata_MCF7.h5ad'
logging.info("Processing MCF7 data from %s", fname)
if not fname.exists():
    logging.info("File not found, creating new MCF7 dataset")
    adata_MCF7 = preprocess_adata_subset_type(adata_sciplex, "MCF7")
    sc.write(fname, adata_MCF7)
    logging.info("MCF7 dataset saved to %s", fname)
else: 
    logging.info("Loading existing MCF7 dataset")
    adata_MCF7 = sc.read(fname)
    
# MCF7 LINCS genes
fname = PROJECT_DIR/'datasets'/'adata_MCF7_lincs_genes.h5ad'    
logging.info("Processing MCF7 LINCS genes data from %s", fname)
if not fname.exists():
    logging.info("File not found, creating new MCF7 LINCS genes dataset")
    adata_MCF7_lincs_genes = preprocess_adata_subset_type(adata_sciplex_lincs_genes, "MCF7")
    sc.write(fname, adata_MCF7_lincs_genes)
    logging.info("MCF7 LINCS genes dataset saved to %s", fname)
else: 
    logging.info("Loading existing MCF7 LINCS genes dataset")
    adata_MCF7_lincs_genes = sc.read(fname)

# K562
fname = PROJECT_DIR/'datasets'/'adata_K562.h5ad'
logging.info("Processing K562 data from %s", fname)
if not fname.exists():
    logging.info("File not found, creating new K562 dataset")
    adata_K562 = preprocess_adata_subset_type(adata_sciplex, "K562")
    sc.write(fname, adata_K562)
    logging.info("K562 dataset saved to %s", fname)
else: 
    logging.info("Loading existing K562 dataset")
    adata_K562 = sc.read(fname)

# K562 LINCS genes
fname = PROJECT_DIR/'datasets'/'adata_K562_lincs_genes.h5ad'
logging.info("Processing K562 LINCS genes data from %s", fname)
if not fname.exists():
    logging.info("File not found, creating new K562 LINCS genes dataset")
    adata_K562_lincs_genes = preprocess_adata_subset_type(adata_sciplex_lincs_genes, "K562")
    sc.write(fname, adata_K562_lincs_genes)
    logging.info("K562 LINCS genes dataset saved to %s", fname)
else: 
    logging.info("Loading existing K562 LINCS genes dataset")
    adata_K562_lincs_genes = sc.read(fname)

# A549
fname = PROJECT_DIR/'datasets'/'adata_A549.h5ad'
logging.info("Processing A549 data from %s", fname)
if not fname.exists():    
    logging.info("File not found, creating new A549 dataset")
    adata_A549 = preprocess_adata_subset_type(adata_sciplex, "A549")
    sc.write(fname, adata_A549)
    logging.info("A549 dataset saved to %s", fname)
else: 
    logging.info("Loading existing A549 dataset")
    adata_A549 = sc.read(fname)

# A549 LINCS genes
fname = PROJECT_DIR/'datasets'/'adata_A549_lincs_genes.h5ad'
logging.info("Processing A549 LINCS genes data from %s", fname)
if not fname.exists():    
    logging.info("File not found, creating new A549 LINCS genes dataset")
    adata_A549_lincs_genes = preprocess_adata_subset_type(adata_sciplex_lincs_genes, "A549")
    sc.write(fname, adata_A549_lincs_genes)
    logging.info("A549 LINCS genes dataset saved to %s", fname)
else: 
    logging.info("Loading existing A549 LINCS genes dataset")
    adata_A549_lincs_genes = sc.read(fname)

logging.info("Completed loading/creating all subsetted adata objects")
# -

# ## Plot pathways for different cell lines

pathways = [
#     'Antioxidant', 
    'Apoptotic regulation', 
    'Cell cycle regulation',
    'DNA damage & DNA repair', 
    'Epigenetic regulation',
#     'Focal adhesion signaling', 
    'HIF signaling', 
    'JAK/STAT signaling',
#     'Metabolic regulation', 
#     'Neuronal signaling',
    'Nuclear receptor signaling', 
#     'Other', 
    'PKC signaling',
    'Protein folding & Protein degradation', 
#     'TGF/BMP signaling',
    'Tyrosine kinase signaling', 
#     'Vehicle'
]


# ### LINCS genes

# +
dose = 1e4

fig, ax = plt.subplots(1, 3, figsize=(20, 5))

sc.pl.umap(
    adata_A549_lincs_genes[adata_A549_lincs_genes.obs.dose==dose].copy(), 
    color='pathway_level_1', 
    groups=pathways,
    legend_fontsize='xx-small',
    show=False, 
    ax=ax[0]
)
sc.pl.umap(
    adata_K562_lincs_genes[adata_K562_lincs_genes.obs.dose==dose].copy(), 
    color='pathway_level_1',
    groups=pathways,
    legend_fontsize='xx-small',
    show=False, 
    ax=ax[1]
)
sc.pl.umap(
    adata_MCF7_lincs_genes[adata_MCF7_lincs_genes.obs.dose==dose].copy(), 
    color='pathway_level_1',
    groups=pathways,
    legend_fontsize='xx-small',
    show=False, 
    ax=ax[2]
)
ax[0].get_legend().remove()
ax[1].get_legend().remove()
plt.tight_layout()
# -

# ### All genes

# +
dose = 1e4

fig, ax = plt.subplots(1, 3, figsize=(20, 5))

sc.pl.umap(
    adata_A549[adata_A549.obs.dose==dose].copy(), 
    color='pathway_level_1', 
    groups=pathways,
    legend_fontsize='xx-small',
    show=False, 
    ax=ax[0]
)
sc.pl.umap(
    adata_K562[adata_K562.obs.dose==dose].copy(), 
    color='pathway_level_1',
    groups=pathways,
    legend_fontsize='xx-small',
    show=False, 
    ax=ax[1]
)
sc.pl.umap(
    adata_MCF7[adata_MCF7.obs.dose==dose].copy(), 
    color='pathway_level_1',
    groups=pathways,
    legend_fontsize='xx-small',
    show=False, 
    ax=ax[2]
)
ax[0].get_legend().remove()
ax[1].get_legend().remove()
plt.tight_layout()

# +
dose = 1e4
cond_A549 = adata_A549_lincs_genes.obs.dose==dose
cond_K562 = adata_K562_lincs_genes.obs.dose==dose
cond_MCF7 = adata_MCF7_lincs_genes.obs.dose==dose

cols = 3
rows = len(pathways)
size = 7
fig, ax = plt.subplots(rows, cols, figsize=(5*cols, 3*rows))

for i, pw in enumerate(pathways):
    sc.pl.umap(adata_A549_lincs_genes[cond_A549].copy(), 
               color='pathway_level_1', 
               groups=pw, 
               show=False, 
               ax=ax[i, 0],
               legend_fontsize='xx-small',
               size=size,
               label=None,
              )

    sc.pl.umap(adata_K562_lincs_genes[cond_K562].copy(), 
               color='pathway_level_1', 
               groups=pw, 
               show=False, 
               ax=ax[i, 1],
               legend_fontsize='xx-small',
               size=size,
               label=None,
              )
    sc.pl.umap(adata_MCF7_lincs_genes[cond_MCF7].copy(), 
               color='pathway_level_1', 
               groups=pw, 
               show=False, 
               ax=ax[i, 2],
               legend_fontsize='xx-small',
               size=size
              ) 
    ax[i, 0].get_legend().remove()
    ax[i, 1].get_legend().remove()

plt.tight_layout()
# -

# ## Distribution of pathways for perturbations with maximal dosage

# ### Lincs genes

# +
dose = 1e4
cond_A549 = adata_A549_lincs_genes.obs.dose==dose
cond_K562 = adata_K562_lincs_genes.obs.dose==dose
cond_MCF7 = adata_MCF7_lincs_genes.obs.dose==dose

cols = 3
rows = len(pathways)
size = 7
fig, ax = plt.subplots(rows, cols, figsize=(6*cols, 3*rows))

for i, pw in enumerate(pathways):
    sc.pl.umap(adata_A549_lincs_genes[cond_A549].copy(), 
               color='pathway_level_1', 
               groups=pw, 
               show=False, 
               ax=ax[i, 0],
               legend_fontsize='xx-small',
               size=size
              )
    sc.pl.umap(adata_K562_lincs_genes[cond_K562].copy(), 
               color='pathway_level_1', 
               groups=pw, 
               show=False, 
               ax=ax[i, 1],
               legend_fontsize='xx-small',
               size=size
              )
    sc.pl.umap(adata_MCF7_lincs_genes[cond_MCF7].copy(), 
               color='pathway_level_1', 
               groups=pw, 
               show=False, 
               ax=ax[i, 2],
               legend_fontsize='xx-small',
               size=size
              )
    ax[i, 0].get_legend().remove()
    ax[i, 1].get_legend().remove()

plt.tight_layout()
# -

# ### All genes

# +
dose = 1e4
cond_A549 = adata_A549.obs.dose==dose
cond_K562 = adata_K562.obs.dose==dose
cond_MCF7 = adata_MCF7.obs.dose==dose

cols = 3
rows = len(pathways)
size = 7
fig, ax = plt.subplots(rows, cols, figsize=(6*cols, 3*rows))

for i, pw in enumerate(pathways):
    sc.pl.umap(adata_A549[cond_A549].copy(), 
               color='pathway_level_1', 
               groups=pw, 
               show=False, 
               ax=ax[i, 0],
               legend_fontsize='xx-small',
               size=size
              )
    sc.pl.umap(adata_K562[cond_K562].copy(), 
               color='pathway_level_1', 
               groups=pw, 
               show=False, 
               ax=ax[i, 1],
               legend_fontsize='xx-small',
               size=size
              )
    sc.pl.umap(adata_MCF7[cond_MCF7].copy(), 
               color='pathway_level_1', 
               groups=pw, 
               show=False, 
               ax=ax[i, 2],
               legend_fontsize='xx-small',
               size=size
              )
    ax[i, 0].get_legend().remove()
    ax[i, 1].get_legend().remove()

plt.tight_layout()
# -

# ## Distribution of drugs with maximal dosage

# ### Identifying significant drugs for perturbations
#
# The relevant information is take from Fig.S6 from the [supplement material](https://www.science.org/doi/suppl/10.1126/science.aax6234/suppl_file/aax6234-srivatsan-sm.pdf) of the orignial [paper](https://www.science.org/doi/full/10.1126/science.aax6234)

# +
epigenetic_drugs = [
    'Dacinostat', 
    'Quisinostat', 
    'CUDC-907', 
    'Abexinostat',
    'Panobinostat',
    'Belinostat',
    'Givinostat',
    'Mocetinostat', #no_ood
    'Pracinostat', #no_ood
    'AR-42', 
    'Entinostat', #no_ood
    'Tucidinostat', #no_ood
    'Tacedinaline', #no_ood
    'Trichostatin', 
    'CUDC-101',
    'M344',
    'Resminostat',
]

dna_damage_drugs = [
    'Raltitrexed', #no_ood
    'Pirarubicin', 
]

cell_cycle_drugs = [
    'Epothilone',
    'Patupilone', #no_ood
    'Flavopiridol',
    'Hesperadin',
    'GSK1070916', #no_ood
]

apoptosis_drugs = [
    'JNJ-26854165' #no_ood
]

tyrosine_drugs = [
    'Trametinib', #no_ood
    'TAK-901',  #no_ood
    'Dasatinib' #no_ood
]

protein_drugs = [
    'Alvespimycin',
    'Tanespimycin',
    'Luminespib'
]


# Create list of potential ood_drugs
ood_drugs = ['control']
ood_drugs.extend(epigenetic_drugs)
ood_drugs.extend(dna_damage_drugs)
ood_drugs.extend(cell_cycle_drugs)
ood_drugs.extend(apoptosis_drugs)
ood_drugs.extend(tyrosine_drugs)
ood_drugs.extend(protein_drugs)
# -

# Create pathway dependent colour palette for more informative plotting 

# +
grey_palette = dict(zip(adata_A549.obs.condition.cat.categories.values, 188*['red']))

colours = (
    ['grey']
    + len(epigenetic_drugs)*['008fd5'] 
    + len(dna_damage_drugs)*['fc4f30'] 
    + len(cell_cycle_drugs)*['e5ae38'] 
    + len(apoptosis_drugs)*['6d904f']
    + len(tyrosine_drugs)*['8b8b8b']
    + len(protein_drugs)*['810f7c']
)

palette = dict(zip(ood_drugs, colours))
for drug, colour in palette.items():
    if drug =='control': 
        continue
    if isinstance(colour, str):
        grey_palette[drug] = '#' + colour
# -

# ### LINCS Genes

# +
dose = 1e4
cond_A549 = (adata_A549_lincs_genes.obs.dose==dose) | (adata_A549_lincs_genes.obs.condition == 'control')
cond_K562 = (adata_K562_lincs_genes.obs.dose==dose) | (adata_K562_lincs_genes.obs.condition == 'control')
cond_MCF7 = (adata_MCF7_lincs_genes.obs.dose==dose) | (adata_MCF7_lincs_genes.obs.condition == 'control')

cols = 3
rows = len(ood_drugs)
fig, ax = plt.subplots(rows, cols, figsize=(5*cols, 3*rows))

for i, drug in enumerate(ood_drugs):
    sc.pl.umap(adata_A549_lincs_genes[cond_A549].copy(), 
               color='condition', 
               groups=drug, 
               show=False, 
               ax=ax[i, 0],
               legend_fontsize='xx-small',
               palette=grey_palette,
               size=20
              )
    sc.pl.umap(adata_K562_lincs_genes[cond_K562].copy(), 
               color='condition', 
               groups=drug, 
               show=False, 
               ax=ax[i, 1],
               legend_fontsize='xx-small',
               palette=grey_palette,
               size=20
              )
    sc.pl.umap(adata_MCF7_lincs_genes[cond_MCF7].copy(), 
               color='condition', 
               groups=drug, 
               show=False, 
               ax=ax[i, 2],
               legend_fontsize='xx-small',
               palette=grey_palette,
               size=20
              ) 
    ax[i, 0].get_legend().remove()
    ax[i, 1].get_legend().remove()

plt.tight_layout()

# -

# ### All genes

# +
dose = 1e4
cond_A549 = (adata_A549.obs.dose==dose) | (adata_A549.obs.condition == 'control')
cond_K562 = (adata_K562.obs.dose==dose) | (adata_K562.obs.condition == 'control')
cond_MCF7 = (adata_MCF7.obs.dose==dose) | (adata_MCF7.obs.condition == 'control')

cols = 3
rows = len(ood_drugs)
fig, ax = plt.subplots(rows, cols, figsize=(5*cols, 3*rows))

for i, drug in enumerate(ood_drugs):
    sc.pl.umap(adata_A549[cond_A549].copy(), 
               color='condition', 
               groups=drug, 
               show=False, 
               ax=ax[i, 0],
               legend_fontsize='xx-small',
               palette=grey_palette,
               size=20
              )
    sc.pl.umap(adata_K562[cond_K562].copy(), 
               color='condition', 
               groups=drug, 
               show=False, 
               ax=ax[i, 1],
               legend_fontsize='xx-small',
               palette=grey_palette,
               size=20
              )
    sc.pl.umap(adata_MCF7[cond_MCF7].copy(), 
               color='condition', 
               groups=drug, 
               show=False, 
               ax=ax[i, 2],
               legend_fontsize='xx-small',
               palette=grey_palette,
               size=20
              ) 
    ax[i, 0].get_legend().remove()
    ax[i, 1].get_legend().remove()

plt.tight_layout()
# -

# ## Create data split

# ### Divide into `'train'`, `'test'`, and `'ood'`

# +
validation_drugs = [    
    'Alvespimycin',  
    'Luminespib',
    'Epothilone',
    'Flavopiridol',
    'Quisinostat', 
    'Abexinostat',
    'Panobinostat',
    'AR-42', 
    'Trichostatin', 
    'M344',
    'Resminostat',
    'Belinostat', #ood
    'Mocetinostat', #no_ood
    'Pracinostat', #no_ood
    'Entinostat', #no_ood
    'Tucidinostat', #no_ood
    'Tacedinaline', #no_ood
    'Patupilone', #no_ood
    'GSK1070916', #no_ood
    'JNJ-26854165' #no_ood
    'TAK-901',  #no_ood
    'Dasatinib' #no_ood
]
    
    
ood_drugs = [
    'Dacinostat', #ood
    'CUDC-907', #ood     
    'Givinostat', #ood
    'CUDC-101', #ood
    'Pirarubicin', #ood 
    'Hesperadin', #ood
    'Tanespimycin', #ood
    'Trametinib', #ood
    'Raltitrexed', #no_ood
]
    
additional_validation_drugs = [
    'YM155', #apoptosis
    'Barasertib', #cell cycle
    'Fulvestrant', #nuclear receptor
    'Nintedanib', #tyrosine
    'Rigosertib', #tyrosine 
    'BMS-754807', #tyrosine
    'KW-2449', #tyrosine
    'Crizotinib', #tyrosin
    'ENMD-2076', #cell cycle 
    'Alisertib', #cell cycle
    'JQ1', #epigenetic
]

validation_drugs.extend(additional_validation_drugs)
# -

# ### Plot additonal validation drugs on all genes data

# +
dose = 1e4
cond_A549 = (adata_A549.obs.dose==dose) | (adata_A549.obs.condition == 'control')
cond_K562 = (adata_K562.obs.dose==dose) | (adata_K562.obs.condition == 'control')
cond_MCF7 = (adata_MCF7.obs.dose==dose) | (adata_MCF7.obs.condition == 'control')

cols = 3
rows = len(additional_validation_drugs)
fig, ax = plt.subplots(rows, cols, figsize=(5*cols, 3*rows))

for i, drug in enumerate(additional_validation_drugs):
    sc.pl.umap(adata_A549[cond_A549].copy(), 
               color='condition', 
               groups=drug, 
               show=False, 
               ax=ax[i, 0],
               legend_fontsize='xx-small',
               palette=grey_palette,
               size=20
              )
    sc.pl.umap(adata_K562[cond_K562].copy(), 
               color='condition', 
               groups=drug, 
               show=False, 
               ax=ax[i, 1],
               legend_fontsize='xx-small',
               palette=grey_palette,
               size=20
              )
    sc.pl.umap(adata_MCF7[cond_MCF7].copy(), 
               color='condition', 
               groups=drug, 
               show=False, 
               ax=ax[i, 2],
               legend_fontsize='xx-small',
               palette=grey_palette,
               size=20
              ) 
    ax[i, 0].get_legend().remove()
    ax[i, 1].get_legend().remove()

plt.tight_layout()

# +
# train
adata_sciplex.obs['split_ood_finetuning'] = 'train'

# ood
adata_sciplex.obs.loc[adata_sciplex.obs.condition.isin(ood_drugs), 'split_ood_finetuning'] = 'ood'

# test
validation_cond = (adata_sciplex.obs.condition.isin(validation_drugs)) & (adata_sciplex.obs.dose.isin([1e3, 1e4]))
val_idx = sc.pp.subsample(adata_sciplex[validation_cond], 0.4, copy=True, random_state=42).obs.index
adata_sciplex.obs.loc[val_idx, 'split_ood_finetuning'] = 'test'

validation_cond = (adata_sciplex.obs.condition.isin(validation_drugs)) & (adata_sciplex.obs.dose.isin([1e1, 1e2]))
val_idx = sc.pp.subsample(adata_sciplex[validation_cond], 0.2, copy=True, random_state=42).obs.index
adata_sciplex.obs.loc[val_idx, 'split_ood_finetuning'] = 'test'

validation_cond = (adata_sciplex.obs.split_ood_finetuning == 'train')
val_idx = sc.pp.subsample(adata_sciplex[validation_cond], 0.04, copy=True, random_state=42).obs.index
adata_sciplex.obs.loc[val_idx, 'split_ood_finetuning'] = 'test'

validation_cond = (adata_sciplex.obs.split_ood_finetuning == 'train') & (adata_sciplex.obs.control.isin([1]))
val_idx = sc.pp.subsample(adata_sciplex[validation_cond], 0.05, copy=True, random_state=42).obs.index
adata_sciplex.obs.loc[val_idx, 'split_ood_finetuning'] = 'test'
# -

adata_sciplex.obs.condition.value_counts()

adata_sciplex.obs['split_ood_finetuning'].value_counts()

pd.crosstab(adata_sciplex.obs['split_ood_finetuning'], 
            adata_sciplex.obs['condition'][adata_sciplex.obs['condition'].isin(ood_drugs)])

pd.crosstab(adata_sciplex.obs.loc[adata_sciplex.obs['split_ood_finetuning']=='ood','dose'], 
            adata_sciplex.obs.loc[adata_sciplex.obs['split_ood_finetuning']=='ood','condition'])

pd.crosstab(adata_sciplex.obs['split_ood_finetuning'], 
            adata_sciplex.obs['condition'][adata_sciplex.obs['condition'].isin(validation_drugs)])

pd.crosstab(adata_sciplex.obs.loc[adata_sciplex.obs['split_ood_finetuning']=='test', 'dose'], 
            adata_sciplex.obs.loc[adata_sciplex.obs['split_ood_finetuning']=='test','condition'])

# ## Add epigenetic holdout split

# +
# train
adata_sciplex.obs['split_ood_finetuning'] = 'train'

# ood
adata_sciplex.obs.loc[adata_sciplex.obs.condition.isin(ood_drugs), 'split_ood_finetuning'] = 'ood'


validation_cond = (adata_sciplex.obs.split_ood_finetuning == 'train')
val_idx = sc.pp.subsample(adata_sciplex[validation_cond], 0.04, copy=True, random_state=42).obs.index
adata_sciplex.obs.loc[val_idx, 'split_ood_finetuning'] = 'test'

validation_cond = (adata_sciplex.obs.split_ood_finetuning == 'train') & (adata_sciplex.obs.control.isin([1]))
val_idx = sc.pp.subsample(adata_sciplex[validation_cond], 0.05, copy=True, random_state=42).obs.index
adata_sciplex.obs.loc[val_idx, 'split_ood_finetuning'] = 'test'

# +
from sklearn.model_selection import train_test_split

epigenetic_drugs_all = adata_sciplex.obs.condition[adata_sciplex.obs.pathway_level_1 == "Epigenetic regulation"].unique()

epigenetic_drugs = [
    'Dacinostat', 
    'Quisinostat', 
    'CUDC-907', 
    'Abexinostat',
    'Panobinostat',
    'Belinostat',
    'Givinostat',
    'AR-42', 
    'Trichostatin', 
    'CUDC-101',
    'M344',
    'Resminostat',
    'Entinostat', #no_ood
    'Tucidinostat', #no_ood
    'Tacedinaline', #no_ood
    'Mocetinostat', #no_ood
    'Pracinostat', #no_ood
]

ood_epi_drugs = [
    'Dacinostat', 
    'Quisinostat', 
    'CUDC-907', 
    'Abexinostat',
    'Panobinostat',
    'Belinostat',
    'Givinostat',
    'AR-42',
]

val_epi_drugs = [
    'Trichostatin', 
    'CUDC-101',
    'M344',
    'Resminostat',
    'Entinostat', #no_ood
    'Tucidinostat', #no_ood
    'Tacedinaline', #no_ood
    'Mocetinostat', #no_ood
    'Pracinostat', #no_ood
]    
    
# -

adata_sciplex.obs.condition.isin(epigenetic_drugs_all).sum()

if 'split_ho_epigenetic' not in list(adata_sciplex.obs):
    print("Addig 'split_ho_epigenetic' to 'adata_sciplex.obs'.")
    obs_train, obs_val = train_test_split(adata_sciplex.obs.index, test_size=0.05, random_state=42)
#     obs_val, obs_test = train_test_split(obs_tmp, test_size=0.5)

    adata_sciplex.obs['split_ho_epigenetic'] = 'train'
    adata_sciplex.obs.loc[adata_sciplex.obs.index.isin(obs_val), 'split_ho_epigenetic'] = 'test'
    
    # test
    validation_cond = (adata_sciplex.obs.condition.isin(val_epi_drugs)) & (adata_sciplex.obs.dose.isin([1e3, 1e4]))
    val_idx = sc.pp.subsample(adata_sciplex[validation_cond], 0.4, copy=True, random_state=42).obs.index
    adata_sciplex.obs.loc[val_idx, 'split_ho_epigenetic'] = 'test'
    

    validation_cond = (adata_sciplex.obs.condition.isin(val_epi_drugs)) & (adata_sciplex.obs.dose.isin([1e1, 1e2]))
    val_idx = sc.pp.subsample(adata_sciplex[validation_cond], 0.4, copy=True, random_state=42).obs.index
    adata_sciplex.obs.loc[val_idx, 'split_ho_epigenetic'] = 'test'
    
    validation_cond = adata_sciplex.obs.condition.isin(epigenetic_drugs_all[~epigenetic_drugs_all.isin(epigenetic_drugs)])
    val_idx = sc.pp.subsample(adata_sciplex[validation_cond], 0.4, copy=True, random_state=42).obs.index
    adata_sciplex.obs.loc[val_idx, 'split_ho_epigenetic'] = 'test'
    
    validation_cond = adata_sciplex.obs.control.isin([True])
    val_idx = sc.pp.subsample(adata_sciplex[validation_cond], 0.4, copy=True, random_state=42).obs.index
    adata_sciplex.obs.loc[val_idx, 'split_ho_epigenetic'] = 'test'
    
    adata_sciplex.obs.loc[adata_sciplex.obs.condition.isin(ood_epi_drugs), 'split_ho_epigenetic'] = 'ood'

if 'split_ho_epigenetic_all' not in list(adata_sciplex.obs):
    print("Addig 'split_ho_epigenetic_all' to 'adata_sciplex.obs'.")
    obs_train, obs_val = train_test_split(adata_sciplex.obs.index, test_size=0.05, random_state=42)
#     obs_val, obs_test = train_test_split(obs_tmp, test_size=0.5)

    adata_sciplex.obs['split_ho_epigenetic_all'] = 'train'
    adata_sciplex.obs.loc[adata_sciplex.obs.index.isin(obs_val), 'split_ho_epigenetic_all'] = 'test'
    
    validation_cond = adata_sciplex.obs.condition.isin(epigenetic_drugs_all[~epigenetic_drugs_all.isin(epigenetic_drugs)])
    val_idx = sc.pp.subsample(adata_sciplex[validation_cond], 0.5, copy=True, random_state=42).obs.index
    adata_sciplex.obs.loc[val_idx, 'split_ho_epigenetic_all'] = 'test'
    
    validation_cond = adata_sciplex.obs.control.isin([True])
    val_idx = sc.pp.subsample(adata_sciplex[validation_cond], 0.4, copy=True, random_state=42).obs.index
    adata_sciplex.obs.loc[val_idx, 'split_ho_epigenetic_all'] = 'test'
    
    adata_sciplex.obs.loc[adata_sciplex.obs.condition.isin(epigenetic_drugs), 'split_ho_epigenetic_all'] = 'ood'
    

pd.crosstab(
    adata_sciplex.obs.split_ho_epigenetic,
    adata_sciplex.obs.control
)

pd.crosstab(
    adata_sciplex.obs.split_ho_epigenetic_all,
    adata_sciplex.obs.control
)

# ## Add random split

if 'split_random' not in list(adata_sciplex.obs):
    print("Addig 'split_random' to 'adata_sciplex.obs'.")
    obs_train, obs_tmp = train_test_split(adata_sciplex.obs.index, test_size=0.3, random_state=42)
    obs_val, obs_test = train_test_split(obs_tmp, test_size=0.5, random_state=42)

    adata_sciplex.obs['split_random'] = 'train'
    adata_sciplex.obs.loc[adata_sciplex.obs.index.isin(obs_val), 'split_random'] = 'test'
    adata_sciplex.obs.loc[adata_sciplex.obs.index.isin(obs_test), 'split_random'] = 'ood'

pd.crosstab(
    adata_sciplex.obs.split_random,
    adata_sciplex.obs.control
)

# ## Save `adata_sciplex` and `adata_sciplex_lincs_genes`

# +
assert (adata_sciplex.obs.index == adata_sciplex_lincs_genes.obs.index).all()

adata_sciplex_lincs_genes.obs['split_ood_finetuning'] = adata_sciplex.obs['split_ood_finetuning']
adata_sciplex_lincs_genes.obs['split_ho_epigenetic'] = adata_sciplex.obs['split_ho_epigenetic']
adata_sciplex_lincs_genes.obs['split_ho_epigenetic_all'] = adata_sciplex.obs['split_ho_epigenetic_all']
adata_sciplex_lincs_genes.obs['split_random'] = adata_sciplex.obs['split_random']
# -

sc.write(PROJECT_DIR/'datasets'/'sciplex_complete_v2.h5ad', adata_sciplex)

sc.write(PROJECT_DIR/'datasets'/'sciplex_complete_lincs_genes_v2.h5ad', adata_sciplex_lincs_genes)

# ______

adata_sciplex = sc.read(PROJECT_DIR/'datasets'/'sciplex_complete_v2.h5ad')
adata_sciplex

adata_sciplex_lincs_genes = sc.read(PROJECT_DIR/'datasets'/'sciplex_complete_lincs_genes_v2.h5ad')
adata_sciplex_lincs_genes

# _________

# ### Check splits

pd.crosstab(
    adata_sciplex.obs.split_ood_finetuning,
    adata_sciplex.obs.control
)

pd.crosstab(
    adata_sciplex.obs.split_ho_pathway,
    adata_sciplex.obs.control
)

# ____

# ## Create small sciplex dataset

# +
adatas = []

for perturbation in np.unique(adata_sciplex.obs.condition): 
    tmp = adata_sciplex[adata_sciplex.obs.condition == perturbation].copy()
    tmp = sc.pp.subsample(tmp, n_obs=40, copy=True, random_state=42)
    adatas.append(tmp)

adata_subset = adatas[0].concatenate(adatas[1:])
adata_subset.uns = adata_sciplex.uns.copy()

adata_subset
# -

if 'split' not in list(adata_subset.obs):
    print("Addig 'split' to 'adata_subset.obs'.")
    obs_train, obs_tmp = train_test_split(adata_subset.obs.index, test_size=0.3)
    obs_val, obs_test = train_test_split(obs_tmp, test_size=0.5)

    adata_subset.obs['split'] = 'train'
    adata_subset.obs.loc[adata_subset.obs.index.isin(obs_val), 'split'] = 'test'
    adata_subset.obs.loc[adata_subset.obs.index.isin(obs_test), 'split'] = 'ood'

adata_subset.obs.split.value_counts()

sc.write(PROJECT_DIR/'datasets'/'sciplex_complete_subset_lincs_genes_v2.h5ad', adata_subset)

# ____

# ## Create middle sized sci-Plex subset

delete_idx =[]
for drug, df in adata_sciplex.obs.groupby('condition'): 
    # Low dose
    cond = df.dose.isin([10])
    idx = df[cond].sample(frac=0.6).index
    delete_idx.extend(idx)
    # Small dose
    cond = df.dose.isin([100])
    idx = df[cond].sample(frac=0.5).index
    delete_idx.extend(idx)
    # Middle dose 
    cond = df.dose.isin([1000])
    idx = df[cond].sample(frac=0.3).index
    delete_idx.extend(idx)
    # High dose 
    cond = df.dose.isin([10000])
    idx = df[cond].sample(frac=0.15).index
    delete_idx.extend(idx)

cond = ~pd.Series(adata_sciplex.obs.index).isin(delete_idx)

cond.sum()

cond = cond.to_list()

assert (adata_sciplex.obs.index == adata_sciplex_lincs_genes.obs.index).all()

adata = adata_sciplex[cond].copy()
adata_lincs_genes = adata_sciplex_lincs_genes[cond].copy()

pd.crosstab(
    adata.obs.split_ood_finetuning,
    adata.obs.control
)

pd.crosstab(
    adata.obs.split_ho_pathway,
    adata.obs.control
)

sc.write(PROJECT_DIR/'datasets'/'sciplex_complete_middle_subset_v2.h5ad', adata)

sc.write(PROJECT_DIR/'datasets'/'sciplex_complete_middle_subset_lincs_genes_v2.h5ad', adata_lincs_genes)