-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathannotate.py
1365 lines (1263 loc) · 48 KB
/
annotate.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 modules
import pandas as pd
import os
from pyliftover import LiftOver
import numpy as np
from sys import argv
import sys
import time
import Query_api
import subprocess
import tempfile
# disable pandas helper messages
pd.options.mode.chained_assignment = None # default='warn'
### Define helper functions
# Function to validate build and return a standardized string
def run_function(func, message, *args):
try:
func(*args)
except:
sys.exit(1, message)
return True
def validate_build(build):
if build.upper() == "HG19" or build.upper() == "GRCH37":
build = "GRCH37"
elif build.upper() == "HG38" or build.upper() == "GRCH38":
build = "GRCH38"
else:
sys.exit("Invalid build. Please use build hg19 or hg38")
return build
# Wrapper function that print function runtime
def runtime(func):
def wrapper(*args, **kwargs):
start = time.time()
func(*args, **kwargs)
end = time.time()
print(f"Runtime: {round(end - start, 2)}")
return wrapper
# Annotate a credible set based on variant position within annotation range
def create_positional_mapping_matrix(df1, df2, colnames=["start", "end"], chr=False):
if not chr == False:
if chr == "X":
chr = 23
df1 = df1[df1["chr"] == chr]
if "pos_37" in df2.columns:
poscol = "pos_37"
else:
poscol = "pos"
# Reshape and perform matrix factorization to check whether a SNP is within a genomic range
starts = df1[colnames[0]].values.reshape(-1, 1)
ends = df1[colnames[1]].values.reshape(-1, 1)
positions = df2[poscol].values.reshape(1, -1)
matches = np.logical_and(starts <= positions, ends >= positions)
return matches
def liftover_row(chrom, pos, lo):
newpos = lo.convert_coordinate(f"chr{chrom}", int(pos))
try:
newpos = newpos[0][1]
except:
newpos = np.nan
print(f"\nUnable to liftover position: {chrom}:{pos}\n")
return newpos
# Fuction to liftover a pandas column of genomic locations to different build
def liftover_df(lo, df, incols, outcol):
df[outcol[0]] = df.apply(
lambda row: liftover_row(f"{row[incols[0]]}", int(row[incols[1]]), lo),
axis=1,
)
df = df.dropna(subset=[outcol[0]])
if len(df) == 0:
print("\nNo variants in credible set could be lifted over\n")
return None
return df
### Functions for annotation of credible sets
# Get the genes in a locus based on the FUMA genes file
def get_genes_in_locus(locus_no, path_to_genes_file, baseline_genes):
genes = pd.read_csv(baseline_genes, sep="\t")
if not "GenomicLocus" in genes.columns:
gene_info = pd.read_csv(genes, sep="\t")
relevant_genes = baseline_genes[
baseline_genes["external_gene_name"].isin(genes["gene"])
]
relevant_genes = pd.concat(
[
relevant_genes,
baseline_genes[baseline_genes["symbol"].isin(genes["gene"])],
]
)
return relevant_genes
genes["GenomicLocus"] = genes["GenomicLocus"].astype(str)
genes["GenomicLocus"] = genes.GenomicLocus.str.split(":")
genes = genes.explode("GenomicLocus")
relevant_genes = pd.DataFrame()
locusname = str(locus_no)
genes_in_loc = genes[genes["GenomicLocus"] == locusname]
relevant_genes = pd.concat([relevant_genes, genes_in_loc])
if not "start" in relevant_genes.columns:
gene_info = pd.read_csv(baseline_genes, sep="\t")
relevant_genes = relevant_genes.merge(
gene_info, left_on="gene", right_on="external_gene_name", how="left"
)
relevant_genes = relevant_genes.merge(
gene_info, left_on="gene", right_on="symbol", how="left"
)
return relevant_genes
# Get the genes in locus based on physical location of the locus
def Genes_in_locus_bp_based(locus_no, GenomicRiskLoci, baseline_genes, lo=None):
Locus = GenomicRiskLoci[
GenomicRiskLoci["GenomicLocus"].astype(str) == str(locus_no)
]
start = Locus["start"].iloc[0]
end = Locus["end"].iloc[0]
chrom = Locus["chr"].iloc[0]
if chrom == "X":
chrom = "23"
# liftover to GRCh37 if build is GRCh38, retry slightly larger windows if a coordinate fails to convert
if not lo == False:
converted = False
i = 0
while not converted:
new_start = lo.convert_coordinate(f"chr{chrom}", int(start) + i)
if len(new_start) > 0:
start = new_start[0][1]
converted = True
else:
i += 50000
converted = False
i = 0
while not converted:
new_end = lo.convert_coordinate(f"chr{chrom}", int(end) - i)
if len(new_end) > 0:
end = new_end[0][1]
converted = True
else:
i += 50000
genes = pd.read_csv(baseline_genes, sep="\t", engine="pyarrow")
genes["chromosome_name"] = genes["chromosome_name"].replace(["X", "x"], "23")
genes = genes[
(genes["end_position"] >= start)
& (genes["start_position"] <= end)
& (genes["chromosome_name"].astype(int) == int(chrom))
]
genes = genes.rename(
columns={
"chromosome_name": "chr",
"start_position": "start",
"end_position": "end",
"ensembl_gene_id": "ensg",
"external_gene_name": "symbol",
}
)
return genes
# Check if the genes are true positive training genes
def check_TP_genes(genes_in_locus, TP_genes):
with open(TP_genes) as file:
TP_genes = [line.rstrip() for line in file]
## check if genes in locus symbol or ensgid are in TP_genes and add column to df
genes_in_locus = genes_in_locus[
genes_in_locus["ensg"].isin(TP_genes) | genes_in_locus["symbol"].isin(TP_genes)
]
return genes_in_locus
# Calculate the distances to the TSS and gene body from the credible set
def get_TSS_distances(creds, genes_in_locus, prob_col, build):
distances = []
weighted_distances = []
TSS_distances = []
weighted_TSS_distances = []
poscol = "pos_37" if "pos_37" in creds.columns else "pos"
centroid = sum(creds[poscol] * creds[prob_col]) / sum(creds[prob_col])
most_likely_snp = creds.loc[creds[prob_col].idxmax()][poscol]
for index, row in genes_in_locus.iterrows():
if row["strand"] == "+":
TSS_distances.append(abs(row["start"] - most_likely_snp))
weighted_TSS_distances.append(abs(row["start"] - centroid))
else:
TSS_distances.append(abs(most_likely_snp - row["end"]))
weighted_TSS_distances.append(abs(centroid - row["end"]))
if row["start"] < centroid < row["end"]:
distances.append(0)
weighted_distances.append(0)
else:
distances.append(
min(
abs(row["start"] - most_likely_snp),
abs(row["end"] - most_likely_snp),
)
)
weighted_distances.append(
min(abs(row["start"] - centroid), abs(row["end"] - centroid))
)
genes_in_locus["TSS_distance"] = TSS_distances
genes_in_locus["weighted_TSS_distance"] = weighted_TSS_distances
genes_in_locus["distance"] = distances
genes_in_locus["weighted_distance"] = weighted_distances
return genes_in_locus
# Create a dictionary of mamga tissue type relevance for GTEx eQTLs
def create_relevance_dict(magma_scores):
magma_scores["score"] = magma_scores["P"].rank(ascending=False) / len(
magma_scores["P"]
)
threshold = 0.05 / len(magma_scores["score"])
magma_scores["score"] = magma_scores.apply(
lambda row: row["score"] * 0.5
if row["P"] > 0.05
else row["score"] * 2
if row["P"] <= threshold
else row["score"],
axis=1,
)
magma_scores["VARIABLE"] = magma_scores["VARIABLE"].str.slice(stop=27)
magma_scores_dict = dict(zip(magma_scores["VARIABLE"], magma_scores["score"]))
return magma_scores_dict
# Get the GTEx eQTLs in a locus
def get_GTEX_eQTLs(creds, genes, prob_col, finemapped_gtex, build, magma_scores=False):
# prep data for input build
if build.upper() == "GRCH38":
creds["eqtl_id"] = creds.apply(
lambda row: f"chr{row['chr']}_{row['pos']}_{row['a1']}_{row['a2']}_b38",
axis=1,
)
creds["eqtl_id2"] = creds.apply(
lambda row: f"chr{row['chr']}_{row['pos']}_{row['a2']}_{row['a1']}_b38",
axis=1,
)
header = "eQTL"
else:
creds["eqtl_id"] = creds.apply(
lambda row: f"{row['chr']}_{row['pos']}_{row['a1']}_{row['a2']}_b37", axis=1
)
creds["eqtl_id2"] = creds.apply(
lambda row: f"{row['chr']}_{row['pos']}_{row['a2']}_{row['a1']}_b37", axis=1
)
header = "eQTL_hg37"
# GTEX get CLPP
CLPP_sum = []
CLPP_max = []
tissue_specific_CLPP_sum = []
tissue_specific_CLPP_max = []
for index, row in genes.iterrows():
maxval, sumval = 0, 0
genepath = os.path.join(finemapped_gtex, row["ensg"].split(".")[0] + ".parquet")
if not os.path.isfile(genepath):
CLPP_sum.append(sumval)
CLPP_max.append(maxval)
tissue_specific_CLPP_sum.append(sumval)
tissue_specific_CLPP_max.append(maxval)
continue
eqtls = pd.read_parquet(os.path.join(genepath), engine="fastparquet")
overlap = pd.merge(
creds, eqtls, left_on="eqtl_id", right_on=header, how="inner"
)
overlap = pd.concat(
[
overlap,
pd.merge(
creds, eqtls, left_on="eqtl_id2", right_on=header, how="inner"
),
]
)
if overlap.shape[0] > 0:
sumval = sum(list(overlap["Probability"] * overlap[prob_col]))
maxval = max(list(overlap["Probability"] * overlap[prob_col]))
CLPP_sum.append(sumval)
CLPP_max.append(maxval)
if type(magma_scores) == dict and overlap.shape[0] > 0:
overlap["TISSUE"] = overlap["TISSUE"].str.slice(stop=27)
overlap["scores"] = overlap["TISSUE"].map(magma_scores)
overlap["Probability"] = overlap["Probability"] * overlap["scores"]
sumval = sum(list(overlap["Probability"] * overlap[prob_col]))
maxval = max(list(overlap["Probability"] * overlap[prob_col]))
tissue_specific_CLPP_sum.append(sumval)
tissue_specific_CLPP_max.append(maxval)
else:
tissue_specific_CLPP_sum.append(sumval)
tissue_specific_CLPP_max.append(maxval)
genes["CLPP_GTEX_eQTL_sum"] = CLPP_sum
genes["CLPP_GTEX_eQTL_max"] = CLPP_max
if type(magma_scores) == dict:
genes["CLPP_GTEX_tissue_weighted_eQTL_sum"] = tissue_specific_CLPP_sum
genes["CLPP_GTEX_tissue_weighted_eQTL_max"] = tissue_specific_CLPP_max
return genes
# Get the GTEx eQTLs in a locus
def get_eqtlgen_eQTLs(creds, genes, prob_col, eQTL_dir, build):
chrom = creds.iloc[0]["chr"]
if build.upper() == "GRCH38":
creds["eqtl_id"] = creds.apply(
lambda row: f"chr{row['chr']}_{row['pos']}_{row['a1']}_{row['a2']}_b38",
axis=1,
)
creds["eqtl_id2"] = creds.apply(
lambda row: f"chr{row['chr']}_{row['pos']}_{row['a2']}_{row['a1']}_b38",
axis=1,
)
creds["eqtl_id3"] = creds.apply(
lambda row: f"{row['chr']}_{row['pos']}_X_X_b38", axis=1
)
header = "hg38"
else:
creds["eqtl_id"] = creds.apply(
lambda row: f"{row['chr']}_{row['pos']}_{row['a1']}_{row['a2']}_b37", axis=1
)
creds["eqtl_id2"] = creds.apply(
lambda row: f"{row['chr']}_{row['pos']}_{row['a2']}_{row['a1']}_b37", axis=1
)
creds["eqtl_id3"] = creds.apply(
lambda row: f"{row['chr']}_{row['pos']}_X_X_b37", axis=1
)
header = "hg37"
CLPP_max = []
CLPP_sum = []
for index, row in genes.iterrows():
maxval, sumval = 0, 0
genepath = os.path.join(
f"{eQTL_dir}", f"chr{chrom}", f"{row['ensg'].upper()}.parquet"
)
if not os.path.isfile(genepath):
CLPP_sum.append(sumval)
CLPP_max.append(maxval)
continue
eqtls = pd.read_parquet(os.path.join(genepath), engine="fastparquet")
overlap = pd.merge(
creds, eqtls, left_on="eqtl_id", right_on=header, how="inner"
)
overlap = pd.concat(
[
overlap,
pd.merge(
creds, eqtls, left_on="eqtl_id2", right_on=header, how="inner"
),
]
)
overlap = pd.concat(
[
overlap,
pd.merge(
creds, eqtls, left_on="eqtl_id3", right_on=header, how="inner"
),
]
)
if overlap.shape[0] > 0:
sumval = sum(list(overlap["Causal_Post._Prob."] * overlap[prob_col]))
maxval = max(list(overlap["Causal_Post._Prob."] * overlap[prob_col]))
CLPP_sum.append(sumval)
CLPP_max.append(maxval)
genes["CLPP_eQTLgen_sum"] = CLPP_sum
genes["CLPP_eQTLgen_max"] = CLPP_max
return genes
# get genes from eQTL catalog
def get_QTL_catalog(creds, genes, prob_col, eQTL_catalog, build):
if build.upper() == "GRCH38":
creds["eqtl_id"] = creds.apply(
lambda row: f"chr{row['chr']}_{row['pos']}_{row['a1']}_{row['a2']}",
axis=1,
)
creds["eqtl_id2"] = creds.apply(
lambda row: f"chr{row['chr']}_{row['pos']}_{row['a2']}_{row['a1']}",
axis=1,
)
header = "QTL_b38"
else:
creds["eqtl_id"] = creds.apply(
lambda row: f"chr{row['chr']}_{row['pos']}_{row['a1']}_{row['a2']}", axis=1
)
creds["eqtl_id2"] = creds.apply(
lambda row: f"chr{row['chr']}_{row['pos']}_{row['a2']}_{row['a1']}", axis=1
)
header = "QTL_b37"
for qtl in ["eQTL", "rQTL"]:
QTL_max_CLPP_max = []
QTL_max_CLPP_sum = []
QTL_mean_CLPP_max = []
QTL_mean_CLPP_sum = []
for index, row in genes.iterrows():
maxval, sumval, max2val, sum2val = 0, 0, 0, 0
genepath = os.path.join(
f"{os.path.join(eQTL_catalog, qtl)}", f"{row['ensg'].upper()}.parquet"
)
if os.path.isfile(genepath):
eqtls = pd.read_parquet(genepath, engine="fastparquet")
overlap = pd.merge(
creds, eqtls, left_on="eqtl_id", right_on=header, how="inner"
)
overlap = pd.concat(
[
overlap,
pd.merge(
creds,
eqtls,
left_on="eqtl_id2",
right_on=header,
how="inner",
),
]
)
if overlap.shape[0] > 0:
sumval = sum(list(overlap["max"] * overlap[prob_col]))
maxval = max(list(overlap["max"] * overlap[prob_col]))
max2val = max(list(overlap["mean"] * overlap[prob_col]))
sum2val = sum(list(overlap["mean"] * overlap[prob_col]))
QTL_max_CLPP_max.append(maxval)
QTL_max_CLPP_sum.append(sumval)
QTL_mean_CLPP_max.append(max2val)
QTL_mean_CLPP_sum.append(sum2val)
genes[f"max_CLPP_{qtl}_eQTLCatalog_sum"] = QTL_max_CLPP_sum
genes[f"max_CLPP_{qtl}_eQTLCatalog_max"] = QTL_max_CLPP_max
genes[f"mean_CLPP_{qtl}_eQTLCatalog_sum"] = QTL_mean_CLPP_sum
genes[f"mean_CLPP_{qtl}_eQTLCatalog_max"] = QTL_mean_CLPP_max
return genes
def get_Promoters(prom_dir, genes, creds, prob_col, build):
sums = []
maxs = []
proms = pd.read_parquet(
os.path.join(prom_dir, f"Proms_per_transcript_{build}.parquet")
)
for gene in genes["ensg"]:
prom_df = proms[proms["ensg"] == gene]
if len(prom_df) == 0:
sums.append(0)
maxs.append(0)
continue
## create a matrix that maps which variant is in which captured region
matrix = create_positional_mapping_matrix(prom_df, creds)
matrix2 = pd.DataFrame(np.outer(np.ones(len(prom_df)), creds[prob_col]))
matrix2 = matrix2.where(matrix, other=0)
sums.append(matrix2.values.sum())
maxs.append(matrix2.values.max())
genes["Promoter_sum"] = sums
genes["Promoter_max"] = maxs
return genes
# Get the GTEx eQTLs in a locus
def get_VEP(creds, genes, prob_col, build):
scores = {}
VEP_dict = {"HIGH": 1, "MODERATE": 0.6, "LOW": 0.4, "MODIFIER": 0.1}
VEPs = creds.apply(
lambda row: Query_api.query_VEP(
row["chr"], row["pos"], row["a1"], row["a2"], build
),
axis=1,
)
if any(VEPs.apply(lambda x: isinstance(x, str))):
return "Cancel annotation due to timeout in VEP"
vars = []
found_genes = []
consequences = []
for i, query in enumerate(VEPs):
variants = query[0]
PiP = creds[prob_col][i]
if not "transcript_consequences" in variants.keys():
continue
for variant in variants["transcript_consequences"]:
if not "gene_id" in variant.keys():
continue
consequence = float(PiP) * float(VEP_dict[variant["impact"]])
vars.append(i)
found_genes.append(variant["gene_id"])
consequences.append(consequence)
vep_df = pd.DataFrame()
vep_df["ensg"] = found_genes
vep_df["consequence"] = consequences
vep_df['variant'] = vars
vep_df = vep_df.groupby(["ensg", "variant"])["consequence"].max().reset_index()
vep_df["VEP_max"] = vep_df.groupby("ensg")["consequence"].transform("max")
vep_df["VEP_sum"] = vep_df.groupby("ensg")["consequence"].transform("sum")
vep_df = vep_df[["ensg", "VEP_sum", "VEP_max"]]
genes = genes.merge(vep_df, on="ensg", how="left")
genes.fillna(0, inplace=True)
return genes
# run vep in environment
def run_vep_within_environment(
input_file, output_file, vep_path, vep_cache, build, environment_path
):
if build.upper() == "GRCH38" or build.upper() == "HG38":
build = "GRCh38"
else:
build = "GRCh37"
environment_command = f"{environment_path} "
vep_command = f" {vep_path} -i {input_file} -o {output_file} --cache {vep_cache} --assembly {build} --offline --force_overwrite"
subprocess.run(environment_command + vep_command, shell=True, check=True)
return
def cmd_VEP(
creds, genes, prob_col, build, VEP_path, VEP_cache, environment_path, outdir
):
tmpname = tempfile.mktemp()
tmp_file_path = os.path.join(f"{tmpname}.vcf")
output_file_path = os.path.join(f"{tmpname}_output.vcf")
if not os.path.exists(outdir):
os.makedirs(outdir)
output_lines = creds.copy()
output_lines = output_lines.sort_values(["chr", "pos"])
output_lines[['a1', 'a2']] = output_lines[['a1', 'a2']].apply(lambda x: sorted(x), axis=1, result_type='expand')
output_lines = output_lines.apply(
lambda row: [
f"{row['chr']} {row['pos']} {row['pos']} {row['a1']}/{row['a2']} +",
f"{row['chr']} {row['pos']} {row['pos']} {row['a1']}/{row['a2']} -",
],
axis=1,
)
# Flatten the list of output lines
output_lines = [line for variant_lines in output_lines for line in variant_lines]
# Write the lines to a file
with open(tmp_file_path, "w") as file:
file.write("\n".join(output_lines))
run_vep_within_environment(
tmp_file_path, output_file_path, VEP_path, VEP_cache, build, environment_path
)
data = []
with open(output_file_path, "r") as file:
for line in file:
if not line.startswith("##"):
data.append(line.strip().split("\t"))
vep_df = pd.DataFrame(data[1:], columns=data[0])
vep_df["Extra"] = vep_df["Extra"].str.extract(r"IMPACT=(\w+);")
vep_df["Extra"] = vep_df["Extra"].fillna("None")
VEP_dict = {"HIGH": 1, "MODERATE": 0.6, "LOW": 0.4, "MODIFIER": 0.1, "None": 0}
vep_df["VEP"] = vep_df["Extra"].apply(lambda x: VEP_dict[x])
vep_df = vep_df.groupby(["Gene", "#Uploaded_variation"])["VEP"].max().reset_index()
creds["VEP_merge_id"] = creds.apply(
lambda row: f"{row['chr']}_{row['pos']}_{row['a1']}/{row['a2']}",
axis=1,
)
vep_df = vep_df.merge(
creds, left_on="#Uploaded_variation", right_on="VEP_merge_id", how="left"
)
creds.drop(columns=["VEP_merge_id"], inplace=True)
vep_df["VEP_weighted"] = vep_df["VEP"] * vep_df[prob_col]
vep_df["VEP_max"] = vep_df.groupby("Gene")["VEP_weighted"].transform("max")
vep_df["VEP_sum"] = vep_df.groupby("Gene")["VEP_weighted"].transform("sum")
vep_df = vep_df[["Gene", "VEP_sum", "VEP_max"]]
genes = genes.merge(vep_df, left_on="ensg", right_on="Gene", how="left")
genes.drop(columns=["Gene"], inplace=True)
genes.fillna(0, inplace=True)
for f in [
tmp_file_path,
output_file_path,
output_file_path + "_summary.html",
output_file_path + "_warnings.txt",
]:
if os.path.exists(f):
os.remove(f)
return genes
# Get the CADD scores within each gene in the locus
def get_CADD(creds, genes, prob_col, build):
creds["CADD"] = creds.apply(
lambda row: Query_api.query_CADD(
row["chr"], row["pos"], row["a1"], row["a2"], build
),
axis=1,
)
matrix = create_positional_mapping_matrix(genes, creds)
creds["CADD_weighted"] = creds["CADD"] * creds[prob_col]
genes["CADD_sum"] = np.dot(matrix, creds["CADD_weighted"])
max_values = np.nanmax(np.where(matrix, creds["CADD_weighted"], 0), axis=1)
genes["CADD_max"] = max_values
return genes
def tabix_CADD(chr, pos, a1, a2, CADD_path, tabix_path):
region_of_interest = f"{chr}:{pos}-{pos}"
# Run the tabix command and capture the output
command = f"{tabix_path} {CADD_path} {region_of_interest}"
result = subprocess.run(command, capture_output=True, shell=True)
tabix_output = result.stdout.decode("utf-8")
header = ["chr", "pos", "a1", "a2", "raw", "PHRED"]
tabix_output = tabix_output.split("\n")
tabix_output = [x.split("\t") for x in tabix_output if len(x.strip()) > 0]
df = pd.DataFrame(tabix_output, columns=header) # ,dtype={'p':'string'})
if len(df) == 0:
return 0
ndf = df[(df["a1"] == a1) & (df["a2"] == a2)]
if len(ndf) == 1:
return float(ndf["PHRED"].iloc[0])
ndf = df[(df["a1"] == a2) & (df["a2"] == a1)]
if len(ndf) == 1:
return float(ndf["PHRED"].iloc[0])
return 0
def get_CADD_cmd(creds, genes, prob_col, CADD_path, tabix_path):
creds["CADD"] = creds.apply(
lambda row: tabix_CADD(
row["chr"], row["pos"], row["a1"], row["a2"], CADD_path, tabix_path
),
axis=1,
)
matrix = create_positional_mapping_matrix(genes, creds)
creds["CADD_weighted"] = creds["CADD"] * creds[prob_col]
genes["CADD_sum"] = np.dot(matrix, creds["CADD_weighted"])
max_values = np.nanmax(np.where(matrix, creds["CADD_weighted"], 0), axis=1)
genes["CADD_max"] = max_values
return
# Get the Jung HiC interaction scores for each gene in the locus
def get_jung_HiC(jung, genes, creds, prob_col):
sums = []
maxs = []
for gene in genes["ensg"]:
if not os.path.exists(os.path.join(jung, gene + ".parquet")):
sums.append(0)
maxs.append(0)
continue
jungs = pd.read_parquet(
os.path.join(jung, gene + ".parquet"), engine="fastparquet"
)
## split jungs 'raw' on '.' and split into three columns called chr, start and end
jungs[["tmp", "start", "end"]] = jungs["target"].str.split(".", expand=True)
jungs["p"] = jungs["dist_pvalue"]
jungs["start"] = jungs["start"].astype(int)
jungs["end"] = jungs["end"].astype(int)
jungs = jungs.drop(jungs[jungs["p"] >= 1000].index)
## create a matrix that maps which variant is in which captured region
matrix = create_positional_mapping_matrix(jungs, creds)
matrix2 = pd.DataFrame(np.outer(jungs["p"], creds[prob_col]))
matrix2 = matrix2.where(matrix, other=0)
sums.append(matrix2.values.sum())
maxs.append(matrix2.values.max())
genes["Jung_HiC_sum"] = sums
genes["Jung_HiC_max"] = maxs
return genes
# Get the Javierre HiC interaction scores for each gene in the locus
def get_javierre_HiC(javierre, genes, creds, prob_col):
sums = []
maxs = []
for gene in genes["ensg"]:
if not os.path.exists(os.path.join(javierre, gene + ".parquet")):
sums.append(0.0)
maxs.append(0.0)
continue
javierres = pd.read_parquet(
os.path.join(javierre, gene + ".parquet"), engine="fastparquet"
)
celltypes = [
"Mon",
"Mac0",
"Mac1",
"Mac2",
"Neu",
"MK",
"EP",
"Ery",
"FoeT",
"nCD4",
"tCD4",
"aCD4",
"naCD4",
"nCD8",
"tCD8",
"nB",
"tB",
]
## create a max and sum column in javierres based on the celltypes values
javierres["max"] = javierres[celltypes].max(axis=1)
javierres["sum"] = javierres[celltypes].sum(axis=1)
# melt the dataframe to have each celltype measurement into individual rows
matrix = create_positional_mapping_matrix(
javierres, creds, ["oeStart", "oeEnd"]
)
matrix2 = pd.DataFrame(np.outer(javierres["sum"], creds[prob_col])).where(
matrix, other=0
)
matrix3 = pd.DataFrame(np.outer(javierres["max"], creds[prob_col])).where(
matrix, other=0
)
sums.append(matrix2.values.sum())
maxs.append(matrix3.values.max())
genes["Javierre_HiC_sum"] = sums
genes["Javierre_HiC_max"] = maxs
return genes
# Get the ABC EP interaction scores for each gene in the locus
def get_HACER(HACER, genes, creds, prob_col):
for method in ["CAGE", "PRO-seq_GRO-seq"]:
sums = []
maxs = []
for gene in genes["ensg"]:
HACERs = pd.DataFrame()
for d in ["F5", "closest", "4D"]:
try:
tmpdf = pd.read_parquet(
os.path.join(HACER, method, d, gene + ".parquet"),
engine="fastparquet",
)
HACERs = pd.concat([HACERs, tmpdf])
except:
pass
if len(HACERs) == 0:
sums.append(0.0)
maxs.append(0.0)
continue
HACERs["chr"] = HACERs["chr"].str.replace("chr", "")
if not "density" in HACERs.columns:
HACERs["density"] = 1
HACERs["start"] = HACERs["start"].astype(int)
HACERs["end"] = HACERs["end"].astype(int)
matrix = create_positional_mapping_matrix(HACERs, creds)
matrix2 = pd.DataFrame(np.outer(HACERs["density"], creds[prob_col])).where(
matrix, other=0
)
sums.append(matrix2.values.sum())
maxs.append(matrix2.values.max())
genes[f"HACER_{method}_sum"] = sums
genes[f"HACER_{method}_max"] = maxs
return genes
# Get the ABC EP interaction scores for each gene in the locus
def get_CICERO(cicero, genes, creds, prob_col):
sums = []
maxs = []
for gene in genes["ensg"]:
if not os.path.exists(os.path.join(cicero, gene + ".parquet")):
sums.append(0.0)
maxs.append(0.0)
continue
cics = pd.read_parquet(
os.path.join(cicero, gene + ".parquet"), engine="fastparquet"
)
cics["chr"] = cics["chr"].str.replace("chr", "")
matrix = create_positional_mapping_matrix(cics, creds)
matrix2 = pd.DataFrame(np.outer(cics["cor"], creds[prob_col])).where(
matrix, other=0
)
sums.append(matrix2.values.sum())
maxs.append(matrix2.values.max())
genes[f"cicero_sum"] = sums
genes[f"cicero_max"] = maxs
return genes
# Get the ABC EP interaction scores for each gene in the locus
def get_ABC_EP(ABC_EP, genes, creds, prob_col):
sums = []
maxs = []
for gene in genes["ensg"]:
if not os.path.exists(os.path.join(ABC_EP, gene + ".parquet")):
sums.append(0.0)
maxs.append(0.0)
continue
ABCs = pd.read_parquet(
os.path.join(ABC_EP, gene + ".parquet"), engine="fastparquet"
)
matrix = create_positional_mapping_matrix(ABCs, creds)
matrix2 = pd.DataFrame(np.outer(ABCs["ABC.Score"], creds[prob_col])).where(
matrix, other=0
)
sums.append(matrix2.values.sum())
maxs.append(matrix2.values.max())
genes["ABC_EP_sum"] = sums
genes["ABC_EP_max"] = maxs
return genes
# Get the ABC CRISPR interaction scores for each gene in the locus
def get_ABC_CRISPR(ABC_CRISPR, genes, creds, prob_col):
sums = []
maxs = []
for gene in genes["ensg"]:
if not os.path.exists(os.path.join(ABC_CRISPR, gene + ".parquet")):
sums.append(0.0)
maxs.append(0.0)
continue
ABCs = pd.read_parquet(
os.path.join(ABC_CRISPR, gene + ".parquet"), engine="fastparquet"
)
matrix = create_positional_mapping_matrix(ABCs, creds)
matrix2 = pd.DataFrame(np.outer(ABCs["ABC_max"], creds[prob_col])).where(
matrix, other=0
)
sums.append(matrix2.values.sum())
maxs.append(matrix2.values.max())
genes["ABC_CRISPR_sum"] = sums
genes["ABC_CRISPR_max"] = maxs
return genes
# Get the FANTHOM5 interaction scores for each gene in the locus
def get_F5(F5_ep, genes, creds, prob_col):
sums = []
maxs = []
for gene in genes["ensg"]:
if not os.path.exists(os.path.join(F5_ep, gene + ".parquet")):
sums.append(0.0)
maxs.append(0.0)
continue
F5s = pd.read_parquet(
os.path.join(F5_ep, gene + ".parquet"), engine="fastparquet"
)
F5s[["chr", "start", "end"]] = F5s["enhancer"].str.split(r"[:|-]", expand=True)
F5s["start"] = F5s["start"].astype(int)
F5s["end"] = F5s["end"].astype(int)
matrix = create_positional_mapping_matrix(F5s, creds)
matrix2 = pd.DataFrame(np.outer(F5s["correlation"], creds[prob_col])).where(
matrix, other=0
)
sums.append(matrix2.values.sum())
maxs.append(matrix2.values.max())
genes["F5_ep_sum"] = sums
genes["F5_ep_max"] = maxs
return genes
# Get Genehancer predictions for each gene in locus
def get_Genehancer(genehancer, genes, creds, prob_col):
sums = []
maxs = []
for gene in genes["ensg"]:
if not os.path.exists(os.path.join(genehancer, gene + ".parquet")):
sums.append(0.0)
maxs.append(0.0)
continue
genehancer_scores = pd.read_parquet(
os.path.join(genehancer, gene + ".parquet"), engine="fastparquet"
)
matrix = create_positional_mapping_matrix(genehancer_scores, creds)
matrix2 = pd.DataFrame(
np.outer(genehancer_scores["score"], creds[prob_col])
).where(matrix, other=0)
sums.append(matrix2.values.sum())
maxs.append(matrix2.values.max())
genes["GeneHancer_max"] = maxs
genes["GeneHancer_sum"] = sums
return genes
# Get Roadmap Epigenomics scores for each gene in locus
def get_RoadmapEpi(Roadmap, genes, creds, prob_col):
sums = []
maxs = []
for gene in genes["ensg"]:
if not os.path.exists(os.path.join(Roadmap, gene + ".parquet")):
sums.append(0.0)
maxs.append(0.0)
continue
Epi = pd.read_parquet(
os.path.join(Roadmap, gene + ".parquet"), engine="fastparquet"
)
matrix = create_positional_mapping_matrix(Epi, creds)
matrix2 = pd.DataFrame(np.outer(Epi["max"], creds[prob_col])).where(
matrix, other=0
)
sums.append(matrix2.values.sum())
maxs.append(matrix2.values.max())
genes["RoadmapEpi_sum"] = sums
genes["RoadmapEpi_max"] = maxs
return genes
# Get EpiMap scores for each gene in locus
def get_EpiMap(EpiMap, genes, creds, prob_col):
sums = []
maxs = []
for gene in genes["ensg"]:
if not os.path.exists(os.path.join(EpiMap, gene + ".parquet")):
sums.append(0.0)
maxs.append(0.0)
continue
Epi = pd.read_parquet(
os.path.join(EpiMap, gene + ".parquet"), engine="fastparquet"
)
matrix = create_positional_mapping_matrix(Epi, creds)
matrix2 = pd.DataFrame(np.outer(Epi["corr"], creds[prob_col])).where(
matrix, other=0
)
sums.append(matrix2.values.sum())
maxs.append(matrix2.values.max())
genes["EpiMap_sum"] = sums
genes["EpiMap_max"] = maxs
return genes
# Run all annotation funcs for each annotation dataset
def annotate_credset(
genes,
creds,
build,
Ann_path,
cmd_vep,
vep_cache,
vep_docker,
tabix,
CADD_file,
outdir,
prob_col,
fname,
magma=False,
trainset=False,
filter=False,
):
genes = get_TSS_distances(creds, genes, prob_col, build)
if trainset:
if not check_dist_to_causal(genes, dist=750000):
exit("The true positive causal gene is too far from lead SNP for {fname}")
genes = genes[genes["distance"] <= 750000]
start_time = time.time()
if cmd_vep == False:
genes = get_VEP(creds, genes, prob_col, build)
if len(genes) == 0:
print("\nError in querying VEP, exiting\n")
return genes
else:
genes = cmd_VEP(
creds, genes, prob_col, build, cmd_vep, vep_cache, vep_docker, outdir
)
if len(genes) == 0:
print("\nError in querying VEP, this could be due to the VEP API server. Please try again later or contact our GitHub, exiting\n")
return genes
if type(genes) == str:
print("\nError in querying VEP, exiting\n")
return genes
if CADD_file == False or tabix == False:
genes = get_CADD(creds, genes, prob_col, build)
if len(genes) == 0:
print("\nError in querying CADD, this could be due to the VEP API server. Please try again later or contact our GitHub, exiting\n")
return genes
else:
get_CADD_cmd(creds, genes, prob_col, CADD_file, tabix)
genes = genes.drop_duplicates(subset=["ensg", "symbol"], keep="first")
genes = get_eqtlgen_eQTLs(
creds, genes, prob_col, os.path.join(Ann_path, "eQTLGen"), build
)
genes = genes.drop_duplicates(subset=["ensg", "symbol"], keep="first")
genes = get_GTEX_eQTLs(
creds,
genes,
prob_col,
os.path.join(Ann_path, "GTEx_v8_finemapping_CAVIAR"),
build,
magma,
)
genes = genes.drop_duplicates(subset=["ensg", "symbol"], keep="first")
genes = get_QTL_catalog(
creds, genes, prob_col, os.path.join(Ann_path, "eQTL_catalog"), build
)
genes = genes.drop_duplicates(subset=["ensg", "symbol"], keep="first")
genes = get_HACER(os.path.join(Ann_path, "HACER"), genes, creds, prob_col)
genes = genes.drop_duplicates(subset=["ensg", "symbol"], keep="first")
genes = get_CICERO(
os.path.join(Ann_path, "Cicero_whole_blood"), genes, creds, prob_col
)
genes = genes.drop_duplicates(subset=["ensg", "symbol"], keep="first")
genes = get_jung_HiC(os.path.join(Ann_path, "Jung_PCHiC"), genes, creds, prob_col)
genes = genes.drop_duplicates(subset=["ensg", "symbol"], keep="first")
genes = get_javierre_HiC(
os.path.join(Ann_path, "Javierre_PCHiC"), genes, creds, prob_col
)