-
Notifications
You must be signed in to change notification settings - Fork 0
/
GCDPipe.py
1147 lines (962 loc) · 61.2 KB
/
GCDPipe.py
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
import pandas as pd
import numpy as np
from dash.dependencies import Input, Output, State
from dash import Dash, dcc, html, dash_table
import base64
import io
from sklearn.model_selection import train_test_split
from sklearn.metrics import classification_report, confusion_matrix, roc_curve, roc_auc_score
#from sklearn.model_selection import GridSearchCV
from pactools.grid_search import GridSearchCVProgressBar
import pickle
from sklearn.ensemble import RandomForestClassifier
import shap
import plotly.graph_objects as go
import plotly.express as px
from scipy.stats import mannwhitneyu
import plotly.figure_factory as ff
from sklearn.preprocessing import minmax_scale
#Design
#color_scheme = {
# 'twocat_1':"#07BEB8",
# 'twocat_2':"#8F3985"
# }
#color_scheme = {
# 'twocat_1':"#8F3985",
# 'twocat_2':"#51575A"
# }
#color_scheme = {
# 'twocat_1':"#8F3985",
# 'twocat_2':"#709191"
# }
color_scheme = {
'twocat_1':"#8F3985",
'twocat_2':"#ADB1BA"
}
#Data preprocessing functions
#Gene nomenclature unification
def generate_mapped_tfset(template_genedf_filepath, tfset_filepath, tfset_genecol, tfset_truthcol, tfset_name, mapfile='./app_default_assets/HUGO_mappings.pkl', save=False, savepath=None):
template_df=pd.read_csv(template_genedf_filepath, index_col=0)
tfset_remapped=remap_dataframe(tfset_filepath, dataframe_genecol=tfset_genecol, save=False, toreturn=True, mappingfilepath=mapfile)
tfset_remapped=tfset_remapped.rename(columns={tfset_truthcol:tfset_name})
tfset_remapped=tfset_remapped.reset_index()
tfset_remapped=tfset_remapped[['pipe_genesymbol', tfset_name]]
mapped_tfset=pd.merge(template_df, tfset_remapped, how='left', on='pipe_genesymbol')
mapped_tfset[tfset_name] = mapped_tfset[tfset_name].fillna('No_data')
if save==True:
mapped_tfset.to_csv(savepath, index=False)
return mapped_tfset
def remap_dataframe(dataframe, dataframe_genecol='pipe_genesymbol', leave_only_genenames_and_scorecol=False, scorecol=None, leave_only_genenames=False, drop_old_genecol=True, drop_unmapped=True, aggregate_data_bygene=True, aggregation_function='max', save=False, savepath=None, mappingfilepath='./app_default_assets/HUGO_mappings.pkl', toreturn=True, sep=','):
#dataframe=dataframe.reset_index()
dataframe=dataframe.rename(columns={dataframe_genecol: "pipe_genesymbol_unmapped"}, errors="raise")
dataframe_genecol='pipe_genesymbol_unmapped'
mappingfile=open(mappingfilepath, "rb")
mappings=pickle.load(mappingfile)
genes= dataframe[dataframe_genecol]
map_values=list(mappings.values())
map_values=[item for sublist in map_values for item in sublist]
map_values=np.array(map_values)
mapped_col=[]
for g in genes:
if g in mappings.keys():
mapped_col.append(g)
elif g in map_values:
target_gene=[key for key in mappings if g in mappings[key]]
mapped_col.append(target_gene[0])
else:
mapped_col.append('No_data')
dataframe['pipe_genesymbol']=mapped_col
no_unmapped=dataframe[dataframe['pipe_genesymbol']=='No_data'].shape[0]
print(f"Number of unmapped values: {no_unmapped}")
if drop_old_genecol==True:
dataframe=dataframe.drop(dataframe_genecol, axis=1)
if drop_unmapped==True:
dataframe=dataframe[dataframe['pipe_genesymbol']!='No_data']
if leave_only_genenames==True:
dataframe=dataframe[['pipe_genesymbol']]
if save==True:
dataframe.to_csv(savepath, index=False)
if toreturn==False:
return
else:
return {'data':dataframe, 'no_unmapped':no_unmapped}
elif leave_only_genenames_and_scorecol==True:
dataframe=dataframe[['pipe_genesymbol', scorecol]]
dataframe=dataframe.rename(columns={scorecol:'score'})
if save==True:
dataframe.to_csv(savepath, header=False, index=False)
if toreturn==False:
return
else:
return {'data':dataframe, 'no_unmapped':no_unmapped}
else:
if aggregate_data_bygene==True:
dataframe=dataframe.groupby('pipe_genesymbol').agg(aggregation_function)
dataframe=dataframe.reset_index()
if save==True:
dataframe.to_csv(savepath, sep=sep)
if toreturn==True:
return {'data':dataframe, 'no_unmapped':no_unmapped}
else:
return
#Training-testing set generation from GWAS loci file
def get_gene_loci_df(true_gene_df, gene_col='pipe_genesymbol', leading_variant_col='rsid', leading_variant_coord_col='location', leading_variant_chr_col='chromosome', window_bases=500000,genelocs_gencode_withcoords_filepath='./app_default_assets/genelocs_gencode_withcoords.csv'):
print('Generating the TF set...')
genelocs_gencode_withcoords=pd.read_csv(genelocs_gencode_withcoords_filepath)
genelocs_gencode_withcoords['chromosome']=list(map(lambda x:x.split('hr')[1], genelocs_gencode_withcoords['seqname']))
true_gene_df=true_gene_df.rename(columns={leading_variant_col:'rsid', leading_variant_coord_col:'location',gene_col:'genesymbol', leading_variant_chr_col:'chromosome'})
true_gene_df=true_gene_df[['rsid','location','genesymbol','chromosome']]
true_gene_df['is_True']=1
true_genes=list(true_gene_df['genesymbol'])
step=int(window_bases/2) #window/2 from each side of the leading variant
dfs=[]
for i in range(len(true_gene_df['rsid'])):
leading_varname=true_gene_df['rsid'][i]
leading_var_chrom=true_gene_df['chromosome'][i]
leading_var_bp=true_gene_df['location'][i]
locus_chrom=leading_var_chrom
locus_minbp=leading_var_bp-step
locus_maxbp=leading_var_bp+step
locus=genelocs_gencode_withcoords[(genelocs_gencode_withcoords['chromosome']==str(locus_chrom)) & (((genelocs_gencode_withcoords['start']<=locus_maxbp) & (genelocs_gencode_withcoords['start']>=locus_minbp))|((genelocs_gencode_withcoords['end']<=locus_maxbp) & (genelocs_gencode_withcoords['end']>=locus_minbp)))]
locus['leading_variant']=leading_varname
locus['locus_start']=locus_minbp
locus['locus_end']=locus_maxbp
locus=locus[['genesymbol', 'leading_variant','locus_start', 'locus_end']]
dfs.append(locus)
result=pd.concat(dfs)
result=pd.merge(result,true_gene_df, on='genesymbol', how='left')
result=result.fillna(0)
#print(f'There are {np.sum(result["is_True"])} true genes in the set')
result=result.drop_duplicates('genesymbol')
result.is_True=[1 if x in true_genes else 0 for x in list(result.genesymbol)]
result=result.rename(columns={'genesymbol':'pipe_genesymbol'})
print(f'There are {np.sum(result["is_True"])} true genes in the set')
print('TF set is generated.')
return {'tfset': result,
'risk gene count': np.sum(result["is_True"]),
'total gene count': len(result['pipe_genesymbol'])}
#App code
app = Dash()
app.config['suppress_callback_exceptions'] = True
app.layout = html.Div([
dcc.Store(id='gene_data_unmapped'),
dcc.Store(id='feature_data_unmapped'),
dcc.Store(id='drug_data_unmapped'),
dcc.Store(id='gene_data'),
dcc.Store(id='feature_data'),
dcc.Store(id='drug_data'),
dcc.Store(id='drug_list'),
dcc.Markdown(children='''# GCDPipe: gene, cell type, and drug prioritization for complex traits'''),
dcc.Markdown(children='''The pipeline will use GWAS genetic fine-mapping data and expression profiles across cell types/tissues or other categories of interest to train a Random Forest classifier to distinguish the genes belonging to the risk class. Expression profiles will be ranked by correlation between their SHAP and gene expression values. Receptors for small molecule ligands presented in Cellinker database can be subsetted with the corresponding ligand information. Optionally, drugs can be ranked by maximal probabilities of their targets to be assigned to a risk class.'''),
dcc.Markdown(children='''A list of drugs can also be provided to compare probabilities of their targets to be assigned to a risk class with that for other genes as well as to compare maximal risk probabilities of their targets with those for other drugs.'''),
dcc.Markdown(children='''### Required input files:'''),
dcc.Markdown(children='''1. a .csv file with columns named "pipe_genesymbol" and "is_True", in which the genes derived from GWAS genetic fine-mapping procedure are listed and their risk category (1 or 0) is provided (a True-False set for classifier training). [Example 1 (schizophrenia)] (https://raw.githubusercontent.com/ACDBio/GCDPipe/main/app_input_examples/schizophrenia/schizophrenia_mapped_tf_geneset.csv?token=GHSAT0AAAAAABSVV6MLYUQXLBY3Y6HR75L6YS6ZTVQ), [Example 2 (IBD)] (https://raw.githubusercontent.com/ACDBio/GCDPipe/main/app_input_examples/IBD/IBD_mapped_tf_geneset.csv?token=GHSAT0AAAAAABSVV6MKJOA5YSLCANFNRKKKYS6ZUQA) '''),
dcc.Markdown(children='''> Alternatively, a .csv file can be specified with the columns: "pipe_genesymbol", "rsid", "chromosome", "location" giving information about the leading variants for the genetically fine-mapped loci and the risk genes which are regarded as driving the associations in these loci (GRCh38 variant coordinates should be given). Training-testing gene set will be generated in this case with the genes in the 500 kbase windows around the leading variants. Mark the training-testing set generation and gene symbol unification options for this. [Example (IBD)] (https://raw.githubusercontent.com/ACDBio/GCDPipe/main/app_input_examples/IBD/IBD_unmapped_genetic_finemapping_results.csv?token=GHSAT0AAAAAABSVV6MKOMQTKRTN2EVSB2YYYS6ZVIQ)'''),
dcc.Markdown(children='''2. a .csv file with with columns "pipe_genesymbol" and other columns with custom names - gene expression profiles across tissues/cell types of interest to train the classifier on. [Example 1 (schizophrenia)] (https://raw.githubusercontent.com/ACDBio/GCDPipe/main/app_input_examples/schizophrenia/mapped_ALLEN_GTEx_dataset.csv?token=GHSAT0AAAAAABSVV6MLS5WWAI6JX5WSBUWYYS6ZWHA), [Example 2 (IBD, gzipped)] (https://github.com/ACDBio/GCDPipe/blob/main/app_input_examples/IBD/mapped_DICE_BloodAtlas_Colon_Dropviz_GTEx.csv.gz?raw=true)'''),
dcc.Markdown(children='''### Optional input files (for drug-gene interaction analysis):'''),
dcc.Markdown(children='''3. a .csv file with columns named "DRUGBANK_ID" and "pipe_genesymbol", in which drugs and their gene targets are provided. [Example (is assembled from DrugCentral and DGIDB data; can be used as a default input)] (https://raw.githubusercontent.com/ACDBio/GCDPipe/main/app_input_examples/drug_targets_data.csv?token=GHSAT0AAAAAABSVV6MLQZ2CV6EIH6ST6EVUYS6ZZ7Q)'''),
dcc.Markdown(children='''4. a .csv file with a column named "DRUGBANK_ID" with drugs belinging to the category of interest. [Example 1 (schizophrenia)] (https://raw.githubusercontent.com/ACDBio/GCDPipe/main/app_input_examples/schizophrenia/schizophrenia_drugs.csv?token=GHSAT0AAAAAABSVV6MKTWRL7SVXUHGCRMPAYS6Z4KQ), [Example 2 (IBD)] (https://raw.githubusercontent.com/ACDBio/GCDPipe/main/app_input_examples/IBD/IBD_drugs.csv?token=GHSAT0AAAAAABSVV6MKYBOJR442UN6IA4PWYS6Z3RA)'''),
dcc.Markdown(children='''If Cellinker ligand ranking based on receptor scores is desired, mark the corresponding checkbox.'''),
dcc.Markdown(children='''If the used gene nomenculature is not unified across the files, set an option to unify gene symbols. In this case, they will be remapped to HUGO gene nomenclature. Unmapped genes will be dropped and numeric variables will be grouped by gene symbol summarised with 'max' function.'''),
dcc.Markdown(children='''Note that the provided hyperparameter search space can be changed using the sliders.'''),
dcc.Markdown(children='''See [Github] (https://github.com/ACDBio/GCDPipe) for further explanations.'''),
dcc.Markdown(children='''# Input dataset upload'''),
dcc.Upload(
id='upload-genes',
children=html.Div([
'1. Gene data: Drag and Drop or ',
html.A('Select a File')
]),
style={
'width': '100%',
'height': '60px',
'lineHeight': '60px',
'borderWidth': '1px',
'borderStyle': 'dashed',
'borderRadius': '5px',
'textAlign': 'center',
'margin': '10px'
},
# Allow multiple files to be uploaded
multiple=False
),
dcc.Loading(
id="genes-loading",
children=[html.Div([html.Div(id='upload-genes-display')])],
type="cube",
),
dcc.Checklist(options=[
{'label': 'Locus file is provided: generate the training-testing set from GWAS genetic fine-mapping results', 'value': 'tfset_from_locifile_True'}],
id='tfgen-checklist'),
dcc.Loading(
id="tfgen-loading",
children=[html.Div([html.Div(id='tfgen-display')])],
type="cube",
),
dcc.Upload(
id='upload-features',
children=html.Div([
'2. Feature data (expression profiles): Drag and Drop or ',
html.A('Select a File')
]),
style={
'width': '100%',
'height': '60px',
'lineHeight': '60px',
'borderWidth': '1px',
'borderStyle': 'dashed',
'borderRadius': '5px',
'textAlign': 'center',
'margin': '10px'
},
# Allow multiple files to be uploaded
multiple=False
),
dcc.Loading(
id="features-loading",
children=[html.Div([html.Div(id='upload-features-display')])],
type="cube",
),
dcc.Checklist(options=[
{'label': 'Rank the Cellinker small molecule ligands based on the receptor gene score', 'value': 'rank_cellinker'}],
id='rank_ligands-checklist'),
dcc.Checklist(options=[
{'label': 'Drug-risk gene interaction assessment', 'value': 'drug_analysis_True'}],
id='upload-drugs-checklist'),
html.Div(id='upload-drugs-container'),
dcc.Checklist(options=[
{'label': 'Unify gene nomenculature with HUGO', 'value': 'remap_genes_True'}],
id='remapping-checklist'),
#---
html.Div([
dcc.Upload(
id='upload-drugs',
children=html.Div([
'3. Drug-target interaction data (binary): Drag and Drop or ',
html.A('Select a File')
]),
style={
'width': '100%',
'height': '60px',
'lineHeight': '60px',
'borderWidth': '1px',
'borderStyle': 'dashed',
'borderRadius': '5px',
'textAlign': 'center',
'margin': '10px'
},
# Allow multiple files to be uploaded
multiple=False
),
html.Div(id='upload-drugs-display'),
dcc.Checklist(options=[
{'label': 'Compare drug target gene probabilities to be assigned to the risk class for the selected drugs with that for other genes', 'value': 'drug_target_comparison_True'},
{'label': 'Compare drug selection scores with other drug scores', 'value': 'drug_comparison_True'}],
id='drug-analysis-options-checklist'),
html.Div(id='drug-target-enrichment-options-container')
], id='drug_upload_addons', style={'display':'none'}),
#---
dcc.Loading(
id="remapping-loading",
children=[html.Div([html.Div(id='remapping-display')])],
type="cube",
),
dcc.Markdown(children='''# Training settings'''),
dcc.Markdown(children=''' \n Number of estimators:'''),
dcc.Input(id="input-n_estimators", type="number", placeholder="Enter the number...", style={'marginRight':'100px'}, value=3000),
dcc.Markdown(children=''' \n Testing gene set/training gene set ratio:'''),
dcc.RangeSlider(0, 1, 0.1, value=[0.3], id='input-testing_gs_ratio', tooltip={"placement": "bottom", "always_visible": True}, marks=None),
dcc.Markdown(children=''' \n Max tree depth:'''),
dcc.RangeSlider(1, 100, 1, value=[1], id='input-max_depth', tooltip={"placement": "bottom", "always_visible": True}, marks=None),
dcc.Markdown(children=''' \n Number of samples per leaf (it is possible to set up to 10 values to search from by sliding the points from the default value):'''),
dcc.RangeSlider(1, 50, 1, value=[5,5,5,5,5,5,5,5,5,5], id='input-n_samples_per_leaf', tooltip={"placement": "bottom", "always_visible": True}), #[5,10,15,20,25,30] was used in the original study
dcc.Markdown(children=''' \n Min number of samples required to split an internal node (set up to 10 values to search from by sliding the points from the default value):'''),
dcc.RangeSlider(1, 50, 1, value=[5,5,5,5,5,5,5,5,5,5], id='input-min_samples_split', tooltip={"placement": "bottom", "always_visible": True}), #[2,5,10,20,30,40] was used in the original study
dcc.Markdown(children=''' \n Min impurity decrease (set up to 10 values to search from by sliding the points from the default value):'''),
dcc.RangeSlider(0, 1, 0.05, value=[0,0,0,0,0,0,0,0,0,0], id='input-min_impurity_decrease', tooltip={"placement": "bottom", "always_visible": True}), #0 was used in the original stosy
html.Div([html.Button("Launch the pipeline", id="start_training", style={"padding": "1rem 1rem", "margin-top": "2rem", "margin-bottom": "1rem"}),
dcc.Loading(
id="training-loading",
children=[html.Div([html.Div(id="training-output")])],
type="cube",
)
]
),
])
@app.callback(Output('gene_data_unmapped', 'data'),
Output('upload-genes-display', 'children'),
Output('tfgen-display', 'children'),
Input('upload-genes', 'contents'),
State('upload-genes', 'filename'),
State('upload-genes', 'last_modified'),
Input('tfgen-checklist', 'value'),
prevent_initial_call=True)
def process_gene_input(contents, name, date, tfgen_option):
if contents is not None:
data=parse_content(contents, name, date)
if data[1]=='ok':
data=data[0]
genedata_df=pd.read_json(data, orient='split')
genedata_df_fordisplay=genedata_df.loc[0:2,:]
if tfgen_option is None:
return data, [html.Div('First rows of the generated set:'), dash_table.DataTable(
genedata_df_fordisplay.to_dict('records'),
[{'name': i, 'id': i} for i in genedata_df_fordisplay.columns],
style_table={'overflowX': 'auto'},
style_cell={
'minWidth': '10px', 'width': '100px', 'maxWidth': '500px',
'overflow': 'hidden',
'textOverflow': 'ellipsis'
},
fill_width=False
)], []
if tfgen_option is not None:
if tfgen_option[0]=='tfset_from_locifile_True':
res=get_gene_loci_df(true_gene_df=genedata_df)
tfdata_df=res['tfset']
risk_gene_count=res['risk gene count']
total_gene_count=res['total gene count']
return tfdata_df.to_json(orient='split'), [html.Div('First rows of the generated set:'), dash_table.DataTable(
genedata_df_fordisplay.to_dict('records'),
[{'name': i, 'id': i} for i in genedata_df_fordisplay.columns],
style_table={'overflowX': 'auto'},
style_cell={
'minWidth': '10px', 'width': '100px', 'maxWidth': '500px',
'overflow': 'hidden',
'textOverflow': 'ellipsis'
},
fill_width=False
)], [html.Div(f'Risk gene count in the obtained training-testing gene set: {risk_gene_count}'),
html.Div(f'Total gene count in the obtained training-testing gene set: {total_gene_count}'),
html.Div('The generated set:'),
dash_table.DataTable(
tfdata_df.to_dict('records'),
[{'name': i, 'id': i} for i in tfdata_df.columns],
style_table={'overflowX': 'auto'},
style_cell={
'minWidth': '10px', 'width': '100px', 'maxWidth': '500px',
'overflow': 'hidden',
'textOverflow': 'ellipsis'
},
fill_width=False,
editable=False,
page_current= 0,
page_size= 5,
filter_action="native",
sort_action="native",
export_format="csv"
)]
if data[1]=='parsing_error':
return [], [html.Div('ERROR PROCESSING THE FILE! The file seems not to be in a .csv format. Upload a .csv file, please.')],[]
else:
return [],[],[]
@app.callback(Output('gene_data', 'data'),
Output('feature_data', 'data'),
Output('drug_data', 'data'),
Output('remapping-display', 'children'),
Input('gene_data_unmapped', 'data'),
Input('feature_data_unmapped', 'data'),
Input('drug_data_unmapped', 'data'),
Input('remapping-checklist','value'),
prevent_initial_call=True)
def unify_gene_nomanclature(genedat, featuredat, drugdat, remapping_option):
if remapping_option is not None:
genedat_df=pd.read_json(genedat, orient='split')
featuredat_df=pd.read_json(featuredat, orient='split')
genedat_df=genedat_df[['pipe_genesymbol','is_True']]
gdat_res=remap_dataframe(genedat_df)
genedat_df_mapped=gdat_res['data']
genedat_no_unmapped=gdat_res['no_unmapped']
ftdat_res=remap_dataframe(featuredat_df)
featuredat_df_mapped=ftdat_res['data']
featuredat_no_unmapped=ftdat_res['no_unmapped']
displayoutput=[html.Div(f'Unmapped gene count for gene data: {genedat_no_unmapped}'),
html.Div('Mapped gene data:'),
dash_table.DataTable(
genedat_df_mapped.to_dict('records'),
[{'name': i, 'id': i} for i in genedat_df_mapped.columns],
style_table={'overflowX': 'auto'},
style_cell={
'minWidth': '10px', 'width': '100px', 'maxWidth': '500px',
'overflow': 'hidden',
'textOverflow': 'ellipsis'
},
fill_width=False,
editable=False,
page_current= 0,
page_size= 5,
filter_action="native",
sort_action="native",
export_format="csv"
),
html.Div(f'Unmapped gene count for feature data: {featuredat_no_unmapped}'),
html.Div('Mapped feature data:'),
dash_table.DataTable(
featuredat_df_mapped.to_dict('records'),
[{'name': i, 'id': i} for i in featuredat_df_mapped.columns],
style_table={'overflowX': 'auto'},
style_cell={
'minWidth': '10px', 'width': '100px', 'maxWidth': '500px',
'overflow': 'hidden',
'textOverflow': 'ellipsis'
},
fill_width=False,
editable=False,
page_current= 0,
page_size= 5,
filter_action="native",
sort_action="native",
export_format="csv"
)
]
if drugdat is not None:
drugdat_df=pd.read_json(drugdat, orient='split')
ddat_res=remap_dataframe(drugdat_df)
drugdat_df_mapped=ddat_res['data']
drugdat_no_unmapped=ddat_res['no_unmapped']
displayoutput.append(html.Div(f'Unmapped gene count for drug data: {drugdat_no_unmapped}'))
displayoutput.append(html.Div('Mapped drug data:'))
displayoutput.append(dash_table.DataTable(
drugdat_df_mapped.to_dict('records'),
[{'name': i, 'id': i} for i in drugdat_df_mapped.columns],
style_table={'overflowX': 'auto'},
style_cell={
'minWidth': '10px', 'width': '100px', 'maxWidth': '500px',
'overflow': 'hidden',
'textOverflow': 'ellipsis'
},
fill_width=False,
editable=False,
page_current= 0,
page_size= 5,
filter_action="native",
sort_action="native",
export_format="csv"
))
return genedat_df_mapped.to_json(orient='split'), featuredat_df_mapped.to_json(orient='split'), drugdat_df_mapped.to_json(orient='split'), displayoutput
else:
return genedat_df_mapped.to_json(orient='split'), featuredat_df_mapped.to_json(orient='split'), [], displayoutput
else:
return genedat, featuredat, drugdat, []
def parse_content(contents, filename, date):
content_type, content_string = contents.split(',')
decoded = base64.b64decode(content_string)
try:
if 'csv' in filename:
# Assume that the user uploaded a CSV file
df = pd.read_csv(
io.StringIO(decoded.decode('utf-8')))
elif 'xls' in filename:
# Assume that the user uploaded an excel file
df = [pd.read_excel(io.BytesIO(decoded))]
return [df.to_json(orient='split'), 'ok']
except Exception as e:
print(e)
return [html.Div([
'There was an error processing this file.'
]), 'parsing_error']
@app.callback(Output('feature_data_unmapped', 'data'),
Output('upload-features-display', 'children'),
Input('upload-features', 'contents'),
State('upload-features', 'filename'),
State('upload-features', 'last_modified'),
prevent_initial_call=True)
def update_output(contents, name, date):
if contents is not None:
data=parse_content(contents, name, date)
if data[1]=='ok':
data=data[0]
featuredata_df=pd.read_json(data, orient='split')
featuredata_df_fordisplay=featuredata_df.loc[0:2,:]
return data, [
html.Div('First rows read:'),
dash_table.DataTable(
featuredata_df_fordisplay.to_dict('records'),
[{'name': i, 'id': i} for i in featuredata_df_fordisplay.columns],
style_table={'overflowX': 'auto'},
style_cell={
'height': 'auto',
'minWidth': '180px', 'width': '180px', 'maxWidth': '180px',
'whiteSpace': 'normal'
},
fill_width=False
)]
if data[1]=='parsing_error':
return [], [html.Div('ERROR PROCESSING THE FILE! The file seems not to be in a .csv format. Upload a .csv file, please.')]
else:
return [],[]
@app.callback(Output('upload-drug-list-display', 'children'),
Input('drug_list', 'data'),
prevent_initial_call=True)
def update_output(data):
if data is not None:
if data!='parsing_error':
df=pd.read_json(data, orient='split')
df=df.loc[0:2,:]
return [
html.Div('First rows read:'),
dash_table.DataTable(
df.to_dict('records'),
[{'name': i, 'id': i} for i in df.columns],
style_table={'overflowX': 'auto'},
style_cell={
'height': 'auto',
'minWidth': '180px', 'width': '180px', 'maxWidth': '180px',
'whiteSpace': 'normal'
},
fill_width=False
)]
if data=='parsing_error':
return [html.Div('ERROR PROCESSING THE FILE! The file seems not to be in a .csv format. Upload a .csv file, please.')]
else:
return []
@app.callback(Output('drug_data_unmapped', 'data'),
Output('upload-drugs-display', 'children'),
Input('upload-drugs', 'contents'),
State('upload-drugs', 'filename'),
State('upload-drugs', 'last_modified'))
def update_output(contents, name, date):
if contents is not None:
data_content=parse_content(contents, name, date)
if data_content[1]=='ok':
data=data_content[0]
data = pd.read_json(data, orient='split')
data_fordisplay=data.loc[0:2,:]
return data_content[0], [
html.Div('First rows read:'),
dash_table.DataTable(
data_fordisplay.to_dict('records'),
[{'name': i, 'id': i} for i in data_fordisplay.columns],
style_table={'overflowX': 'auto'},
style_cell={
'height': 'auto',
'minWidth': '180px', 'width': '180px', 'maxWidth': '180px',
'whiteSpace': 'normal'
},
fill_width=False
)]
if data_content[1]=='parsing_error':
return [], [html.Div('ERROR PROCESSING THE FILE! The file seems not to be in a .csv format. Upload a .csv file, please.')]
else:
return [],[]
@app.callback(Output('drug_list', 'data'),
Input('upload-drug-list', 'contents'),
State('upload-drug-list', 'filename'),
State('upload-drug-list', 'last_modified'))
def update_output(contents, name, date):
if contents is not None:
data_content=parse_content(contents, name, date)
if data_content[1]=='ok':
data_content=data_content[0]
if data_content[1]=='parsing_error':
data_content=data_content[1]
return data_content
@app.callback(Output('upload-drugs-container', 'children'),
Input('upload-drugs-checklist', 'value'),
prevent_initial_call=True)
def add_drugs_upload(drug_option_value):
if (drug_option_value is not None) and (drug_option_value!=[]):
drug_option_value=drug_option_value[0]
if drug_option_value=='drug_analysis_True':
return [
dcc.Upload(
id='upload-drugs',
children=html.Div([
'3. Drug-target interaction data (binary): Drag and Drop or ',
html.A('Select a File')
]),
style={
'width': '100%',
'height': '60px',
'lineHeight': '60px',
'borderWidth': '1px',
'borderStyle': 'dashed',
'borderRadius': '5px',
'textAlign': 'center',
'margin': '10px'
},
# Allow multiple files to be uploaded
multiple=False
),
html.Div(id='upload-drugs-display'),
dcc.Checklist(options=[
{'label': 'Compare drug target gene probabilities to be assigned to the risk class for the selected drugs with that for other genes', 'value': 'drug_target_comparison_True'},
{'label': 'Compare drug selection scores with other drug scores', 'value': 'drug_comparison_True'}],
id='drug-analysis-options-checklist'),
html.Div(id='drug-target-enrichment-options-container')
]
@app.callback(Output('drug-target-enrichment-options-container', 'children'),
Input('drug-analysis-options-checklist', 'value'))
def add_drug_list_upload(drug_analysis_options_value):
if drug_analysis_options_value is not None:
if ('drug_target_comparison_True' in drug_analysis_options_value) or ('drug_comparison_True' in drug_analysis_options_value):
return [
dcc.Upload(
id='upload-drug-list',
children=html.Div([
'4. Target drug category Drugbank IDs: Drag and Drop or ',
html.A('Select a File')
]),
style={
'width': '100%',
'height': '60px',
'lineHeight': '60px',
'borderWidth': '1px',
'borderStyle': 'dashed',
'borderRadius': '5px',
'textAlign': 'center',
'margin': '10px'
},
# Allow multiple files to be uploaded
multiple=False
),
html.Div(id='upload-drug-list-display'),
]
else:
return []
@app.callback(Output("training-output", 'children'),
Input('start_training','n_clicks'),
Input('gene_data', 'data'),
Input('feature_data', 'data'),
Input('input-n_estimators','value'),
Input('input-testing_gs_ratio', 'value'),
Input('input-max_depth','value'),
Input('input-n_samples_per_leaf','value'),
Input('input-min_samples_split','value'),
Input('input-min_impurity_decrease','value'),
Input('upload-drugs-checklist','value'),
Input('drug_data', 'data'),
Input('drug-analysis-options-checklist', 'value'),
Input('drug_list', 'data'),
Input('rank_ligands-checklist','value'),
prevent_initial_call=True)
def train_rf_classifier(n_clicks, gene_data, feature_data, n_estimators, test_ratio, max_depth, n_samples_leaf, min_samples_split, min_impurity_decrease, drug_analysis_requirement, d_data, drug_analysis_options, d_list, rank_ligands_requirement='None', cellinker_filepath='./app_default_assets/human-sMOL_remapped.txt'):
if n_clicks is not None:
if n_clicks>0:
if gene_data is None:
return html.Div(['Gene data is missing...'])
if feature_data is None:
return html.Div(['Feature data (gene expression profiles) is missing...'])
gene_df=pd.read_json(gene_data, orient='split')
gene_df=gene_df[['pipe_genesymbol','is_True']]
feature_df=pd.read_json(feature_data, orient='split')
processed_df=pd.merge(feature_df, gene_df, on='pipe_genesymbol',how='left')
processed_df.is_True=processed_df.is_True.fillna('No_data')
classifier_building_df=processed_df[processed_df.is_True!='No_data']
classifier_building_df.is_True=[1 if x==1.0 else 0 for x in classifier_building_df.is_True]
print('True gene count:')
print(classifier_building_df.is_True.sum())
classifier_building_df=classifier_building_df.set_index('pipe_genesymbol')
X_train, X_test, Y_train, Y_test = train_test_split(classifier_building_df.iloc[:,:-1],classifier_building_df.is_True, test_size=test_ratio[0], stratify=classifier_building_df.is_True)
print('Training risk gene count:')
print(Y_train.sum())
print('Training risk gene count:')
print(Y_test.sum())
classifier_tuned_parameters_rf={'bootstrap':[True],
'max_depth':np.unique(max_depth),
'max_features':['auto', 'sqrt'],
'min_samples_leaf':np.unique(n_samples_leaf),
'min_samples_split':np.unique(min_samples_split),
'max_features':[None],
'n_estimators':[n_estimators],
'min_impurity_decrease':np.unique(min_impurity_decrease)}
clf=GridSearchCVProgressBar(RandomForestClassifier(oob_score=True,bootstrap = True), classifier_tuned_parameters_rf, scoring='f1_weighted', verbose=1)
print('Hyperparameter tuning for the rf model...')
clf.fit(X_train, Y_train)
best_fitting_score=clf.best_score_
Y_true, Y_pred = Y_test, clf.best_estimator_.predict_proba(X_test)
print('Y_true')
print(Y_true)
print('Y_pred')
print(Y_pred)
Y_pred=Y_pred[:,1]
fpr, tpr, thresholds = roc_curve(Y_true, Y_pred)
roc_auc=roc_auc_score(Y_true, Y_pred)
optimal_idx = np.argmax(tpr - fpr)
optimal_threshold = thresholds[optimal_idx]
print("Threshold value is:", optimal_threshold)
predicted_classes=[1 if x>optimal_threshold else 0 for x in Y_pred] #here might be reasonable to try just >
cm = confusion_matrix(Y_true, predicted_classes, labels=[0, 1])
cls_report=classification_report(Y_true, predicted_classes, output_dict=True)
print(cls_report)
print(cm)
#precision=cm[1,1]/(cm[0,1]+cm[1,1])
#accuracy=(cm[1,1]+cm[0,0])/(cm[1,1]+cm[0,0]+cm[0,1]+cm[1,0])
#print('Plotting the classifier training stats...')
# Confusion matrix hitmap
#plt.figure()
#sns.heatmap(cm, annot=True, cmap="YlGnBu")
#plt.title(f'Confusion matrix for the optimal ROC-AUC threshold.')
#fig_cm_sns = plt.gcf()
#plt.close()
#ROC-AUC curve
#plt.figure()
#lw = 2
#plt.plot(
# fpr,
# tpr,
# color="darkorange",
# lw=lw,
# label="ROC curve (area = %0.2f)" % roc_auc,
# )
#plt.plot([0, 1], [0, 1], color="navy", lw=lw, linestyle="--")
#plt.xlim([0.0, 1.0])
#plt.ylim([0.0, 1.05])
#plt.xlabel("False Positive Rate")
#plt.ylabel("True Positive Rate")
#plt.title("Receiver operating characteristic")
#plt.legend(loc="lower right")
#fig_roc_sns = plt.gcf()
#plt.close()
fig_roc = go.Figure()
name = f"AUC={roc_auc:.2f}"
fig_roc.add_shape(
type='line', line=dict(dash='dash'),
x0=0, x1=1, y0=0, y1=1
)
fig_roc.add_trace(go.Scatter(x=fpr, y=tpr, mode='lines'))
fig_roc.update_layout(
template='plotly_white',
title=name,
xaxis_title='False Positive Rate',
yaxis_title='True Positive Rate',
yaxis=dict(scaleanchor="x", scaleratio=1),
xaxis=dict(constrain='domain'),
width=500, height=500
)
print('Scoring the whole dataset...')
processed_df=processed_df.set_index('pipe_genesymbol')
processed_df=processed_df.iloc[:,:-1]
predicted_probs=pd.DataFrame.from_dict({'pipe_genesymbol':processed_df.index.tolist()})
predicted_probs['score']=clf.best_estimator_.predict_proba(processed_df)[:,1]
predicted_probs['is_risk_class']=predicted_probs['score']>optimal_threshold
input_risk_genes=list(classifier_building_df[classifier_building_df.is_True==1].index.tolist())
predicted_probs['is_input_risk_gene']=[1 if gene in input_risk_genes else 0 for gene in predicted_probs['pipe_genesymbol']]
predicted_probs=predicted_probs.sort_values(ascending=False, by=['score'])
explainer = shap.TreeExplainer(clf.best_estimator_)
shap_values = explainer.shap_values(processed_df)
shap_riskclass_df_notabs=pd.DataFrame(shap_values[1])
shap_riskclass_df=pd.DataFrame(abs(shap_values[1]))
shap_riskclass_df.columns=processed_df.columns
shap_riskclass_df_notabs.columns=processed_df.columns
shap_riskclass_means=pd.DataFrame(shap_riskclass_df.mean(axis=0)).reset_index().rename(columns={'index':'expression_profile', 0:'mean_abs_shap_riskclass'})
corr_between_shap_and_exprs_values=shap_riskclass_df_notabs.corrwith(processed_df.reset_index(drop=True),axis=0, method='pearson')
riskclass_shap_featurevalue_correlation=corr_between_shap_and_exprs_values.sort_values(ascending=False)
riskclass_shap_featurevalue_correlation=riskclass_shap_featurevalue_correlation.fillna(0)
riskclass_shap_featurevalue_correlation=pd.DataFrame.from_dict({'expression_profile':riskclass_shap_featurevalue_correlation.index.tolist(),'corrcoef_riskclass':list(riskclass_shap_featurevalue_correlation)})
feature_importance_resdf=pd.merge(shap_riskclass_means, riskclass_shap_featurevalue_correlation, on='expression_profile',how='left')
correlation_sign=np.sign(feature_importance_resdf['corrcoef_riskclass'])
feature_importance_resdf['feature_score']=correlation_sign*feature_importance_resdf['mean_abs_shap_riskclass']
feature_importance_resdf=feature_importance_resdf.sort_values('feature_score', ascending=False)
#feature_importance_resdf['mean_shap_value_scaled']= minmax_scale(feature_importance_resdf[['mean_shap_value']], feature_range=(0,1))
res= [
dcc.Markdown(children='''# Classifier test performance'''),
dcc.Graph(figure=fig_roc),
dcc.Markdown(children='''## Classification report'''),
dcc.Markdown(children='''### Non-risk class: '''),
html.Div([f"Precision {round(cls_report['0']['precision'],3)}"]),
html.Div([f"Recall {round(cls_report['0']['recall'],3)}"]),
html.Div([f"F1 score {round(cls_report['0']['f1-score'],3)}"]),
html.Div([f"Support {cls_report['0']['support']}"]),
dcc.Markdown(children='''### Risk class: '''),
html.Div([f"Precision {round(cls_report['1']['precision'], 3)}"]),
html.Div([f"Recall {round(cls_report['1']['recall'], 3)}"]),
html.Div([f"F1 score {round(cls_report['1']['f1-score'], 3)}"]),
html.Div([f"Support {cls_report['1']['support']}"]),
dcc.Markdown(children='''### Global: '''),
html.Div([f"Accuracy {round(cls_report['accuracy'], 3)}"]),
html.Div([f"Macro average precision {round(cls_report['macro avg']['precision'], 3)}"]),
html.Div([f"Macro average recall {round(cls_report['macro avg']['recall'], 3)}"]),
html.Div([f"Macro average f1 score {round(cls_report['macro avg']['f1-score'], 3)}"]),
html.Div([f"Macro average support {cls_report['macro avg']['support']}"]),
html.Div([f" "]),
html.Div([f"Optimal decision threshold {optimal_threshold}"]),
#html.Div([f"Confusion matrix {cm}"]),
dcc.Store(id='gene_probabilities', data=predicted_probs.to_json(orient='split')),
dcc.Store(id='expression_profile_importances', data=feature_importance_resdf.to_json(orient='split')),
dcc.Markdown(children='''# Classification results'''),
dcc.Markdown(children='''### Gene probabilities to be assigned to the risk class: '''),
dash_table.DataTable(
id='gene_risk_probabilities',
data=predicted_probs.to_dict('records'),
columns=[{'name': i, 'id': i} for i in predicted_probs.columns],
style_table={'overflowX': 'auto'},
editable=False,
page_current= 0,
page_size= 10,
filter_action="native",
sort_action="native",
export_format="csv",
style_cell={
'minWidth': '10px', 'width': '100px', 'maxWidth': '500px',
'overflow': 'hidden',
'textOverflow': 'ellipsis'
},
fill_width=False
),
dcc.Markdown(children='''### Importance-based expression profile scores: '''),
dash_table.DataTable(
id='expression_profile_importances',
data=feature_importance_resdf.to_dict('records'),
columns=[{'name': i, 'id': i} for i in feature_importance_resdf.columns],
style_table={'overflowX': 'auto'},
editable=False,
page_current= 0,
page_size= 10,
filter_action="native",
sort_action="native",
export_format="csv",
style_cell={
'minWidth': '10px', 'width': '100px', 'maxWidth': '500px',
'overflow': 'hidden',
'textOverflow': 'ellipsis'
},
fill_width=False
)]
if rank_ligands_requirement is not None:
if rank_ligands_requirement[0]=='rank_cellinker':
cellinker_db=pd.read_csv(cellinker_filepath, sep='\t')
ligand_df=pd.merge(predicted_probs,cellinker_db, on='pipe_genesymbol', how='left')
ligand_df=ligand_df.fillna(0)
ligand_df=ligand_df.sort_values(by='score', axis=0, ascending=False)
res.append(dcc.Markdown('''### Cellinker receptor gene scores and the associated ligands:'''))
res.append(dash_table.DataTable(
id='ligand_scores',
data=ligand_df.to_dict('records'),
columns=[{'name': i, 'id': i} for i in ligand_df.columns],
style_table={'overflowX': 'auto'},
editable=False,
page_current= 0,
page_size= 10,
filter_action="native",
sort_action="native",
export_format="csv",
style_cell={
'minWidth': '10px', 'width': '100px', 'maxWidth': '500px',
'overflow': 'hidden',
'textOverflow': 'ellipsis'
},
fill_width=False
))
if drug_analysis_requirement is not None:
if drug_analysis_requirement[0]=='drug_analysis_True':
res.append(dcc.Markdown(children='''# Drug ranking results'''))
res.append(dcc.Markdown(children='''### Drug scores: '''))
if d_data is None:
res.append(html.Div(['Drug data is missing...']))
else:
drugdata_df=pd.read_json(d_data, orient='split')
drugdata_df=pd.merge(drugdata_df, predicted_probs[['pipe_genesymbol','score']], on='pipe_genesymbol', how='left')
drugdata_df=drugdata_df.dropna()
drugdata_df_agg=drugdata_df.groupby('DRUGBANK_ID')[['score']].agg([np.max,np.mean,np.min,np.std])
drugdata_df_agg.columns = drugdata_df_agg.columns.get_level_values(1)
drugdata_df_agg=drugdata_df_agg.reset_index()
print(drugdata_df_agg)
drugdata_df_agg=drugdata_df_agg.fillna(0)
drugdata_df_agg=drugdata_df_agg.rename(columns={'amax':'max_targetgene_score', 'mean':'mean_targetgene_score', 'amin':'min_targetgene_score', 'std':'targetgene_score_std'})
drugdata_df_agg['has_risk_class_targets']=drugdata_df_agg['max_targetgene_score']>optimal_threshold
drugdata_df_agg=drugdata_df_agg.sort_values(ascending=False, by=['mean_targetgene_score'])
drugdata_df_agg=drugdata_df_agg.reset_index()