-
Notifications
You must be signed in to change notification settings - Fork 28
/
AmpliconSuite-pipeline.py
executable file
·1104 lines (895 loc) · 47.4 KB
/
AmpliconSuite-pipeline.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
#!/usr/bin/env python
# author: Jens Luebeck (jluebeck [at] ucsd.edu)
import argparse
from datetime import datetime
import json
import logging
import os
import re
import socket
from subprocess import *
import sys
import tarfile
import time
from paalib import check_reference, reduce_fasta
from paalib._version import __ampliconsuitepipeline_version__
PY3_PATH = "python3" # updated by command-line arg if specified
metadata_dict = {} # stores the run metadata (bioinformatic metadata)
sample_info_dict = {} # stores the sample metadata
def run_bwa(ref_fasta, fastqs, outdir, sname, nthreads, samtools, samtools_version):
outname = outdir + sname
logging.info("Output prefix: " + outname)
exts = [".sa", ".amb", ".ann", ".pac", ".bwt"]
indexPresent = True
for i in exts:
if not os.path.exists(ref_fasta + i):
indexPresent = False
logging.info("Could not find " + ref_fasta + i + ", building BWA index from scratch. This could take > 60 minutes")
break
if not indexPresent:
cmd = "bwa index " + ref_fasta
call(cmd, shell=True)
logging.info("Performing alignment and sorting\n")
if samtools_version[0] < 1:
cmd = "{{ bwa mem -K 10000000 -t {} {} {} | {} view -Shu - | {} sort -m 4G -@4 - {}.cs; }} 2>{}_aln_stage.stderr".format(
nthreads, ref_fasta, fastqs, samtools, samtools, outname, outname)
else:
cmd = "{{ bwa mem -K 10000000 -t {} {} {} | {} view -Shu - | {} sort -m 4G -@4 -o {}.cs.bam -; }} 2>{}_aln_stage.stderr".format(
nthreads, ref_fasta, fastqs, samtools, samtools, outname, outname)
logging.info(cmd + "\n")
call(cmd, shell=True)
metadata_dict["bwa_cmd"] = cmd
logging.info("Performing duplicate marking & indexing")
final_bam_name = "{}.cs.rmdup.bam".format(outname)
cmd_list = [samtools, "rmdup", "-s", "{}.cs.bam".format(outname), final_bam_name]
logging.info(" ".join(cmd_list) + "\n")
call(cmd_list)
logging.info("Running samtools index")
cmd_list = [samtools, "index", final_bam_name]
logging.info(" ".join(cmd_list) + "\n")
call(cmd_list)
logging.info("Removing temp BAM\n")
cmd = "rm {}.cs.bam".format(outname)
call(cmd, shell=True)
return final_bam_name, outname + "_aln_stage.stderr"
# This is not currently used by AmpliconSuite-pipeline.
def run_freebayes(ref, bam_file, outdir, sname, nthreads, regions, fb_path=None):
# Freebayes cmd-line args
# -f is fasta
# -r is region to call
logging.info("Running freebayes...")
fb_exec = "freebayes"
if fb_path:
fb_exec = fb_path + "/" + fb_exec
while True:
try:
curr_region_tup = regions.pop()
except IndexError:
break
curr_region_string = curr_region_tup[0] + ":" + curr_region_tup[1]
logging.info(curr_region_string + ". " + str(len(regions)) + " items remaining.")
vcf_file = outdir + sname + "_" + curr_region_tup[0] + "_" + curr_region_tup[2] + ".vcf"
replace_filter_field_func = "awk '{ if (substr($1,1,1) != \"#\" ) { $7 = ($7 == \".\" ? \"PASS\" : $7 ) }} 1 ' OFS=\"\\t\""
cmd = "{} --genotype-qualities --standard-filters --use-best-n-alleles 5 --limit-coverage 25000 \
--strict-vcf -f {} -r {} {} | {} > {}".format(fb_exec, ref, curr_region_string, bam_file,
replace_filter_field_func, vcf_file)
logging.info(cmd)
call(cmd, shell=True)
# gzip the new VCF
call("gzip -f " + vcf_file, shell=True)
# This is not currently used by AmpliconSuite-pipeline.
def merge_and_filter_vcfs(chr_names, vcf_list, outdir, sname):
logging.info("Merging VCFs and zipping...\n")
# collect the vcf files to merge
merged_vcf_file = outdir + sname + "_merged.vcf"
relevant_vcfs = [x for x in vcf_list if any([i in x for i in chr_names])]
chrom_vcf_d = {}
for f in relevant_vcfs:
curr_chrom = f.rsplit(".vcf.gz")[0].rsplit("_")[-2:]
chrom_vcf_d[curr_chrom[0] + curr_chrom[1]] = f
# chr_nums = [x.lstrip("chr") for x in chr_names]
pre_chr_str_names = [str(x) for x in range(1, 23)] + ["X", "Y"]
# sort the elements
# include the header from the first one
if args.ref != "GRCh37" and args.ref != "GRCm38":
sorted_chr_names = ["chr" + str(x) for x in pre_chr_str_names]
cmd = "zcat " + chrom_vcf_d["chrM"] + ''' | awk '$4 != "N"' > ''' + merged_vcf_file
else:
sorted_chr_names = [str(x) for x in pre_chr_str_names]
cmd = "zcat " + chrom_vcf_d["MT"] + ''' | awk '$4 != "N"' > ''' + merged_vcf_file
logging.info(cmd + "\n")
call(cmd, shell=True)
# zcat the rest, grepping out all header lines starting with "#"
logging.debug(sorted_chr_names)
for i in sorted_chr_names:
if i == "chrM" or i == "MT":
continue
cmd_p = "zcat " + chrom_vcf_d[i + "p"] + ''' | grep -v "^#" | awk '$4 != "N"' >> ''' + merged_vcf_file
cmd_q = "zcat " + chrom_vcf_d[i + "q"] + ''' | grep -v "^#" | awk '$4 != "N"' >> ''' + merged_vcf_file
logging.info(cmd_p)
call(cmd_p, shell=True)
logging.info(cmd_q)
call(cmd_q, shell=True)
cmd = "gzip -f " + merged_vcf_file
logging.info(cmd)
call(cmd, shell=True)
return merged_vcf_file + ".gz"
def run_cnvkit(ckpy_path, nthreads, outdir, bamfile, seg_meth='cbs', normal=None, ref_fasta=None, vcf=None):
# CNVkit cmd-line args
# -m wgs: wgs data
# -y: assume chrY present
# -n: create flat reference (cnv baseline)
# -p: number of threads
# -f: reference genome fasta
bamBase = os.path.splitext(os.path.basename(bamfile))[0]
cnvkit_version = Popen([PY3_PATH, ckpy_path, "version"], stdout=PIPE, stderr=PIPE).communicate()[0].rstrip()
try:
cnvkit_version = cnvkit_version.decode('utf-8')
except UnicodeError:
pass
metadata_dict["cnvkit_version"] = cnvkit_version
ckRef = AA_REPO + args.ref + "/" + args.ref + "_cnvkit_filtered_ref.cnn"
if normal and args.ref == "GRCh38_viral":
logging.warning("CNVkit does not properly support matched tumor-normal with viral genomes. Ignoring matched-"
"normal and running in tumor-only mode.\n")
logging.info("Running CNVKit batch\n")
if normal and not args.ref == "GRCh38_viral":
# create a version of the stripped reference
reduce_fasta.reduce_fasta(ref_fasta, ref_genome_size_file, outdir)
base = os.path.basename(ref_fasta) # args.ref is the name, ref is the fasta
stripRefG = outdir + os.path.splitext(base)[0] + "_reduced" + "".join(os.path.splitext(base)[1:])
logging.debug("Stripped reference: " + stripRefG)
cmd = "{} {} batch {} -m wgs --fasta {} -p {} -d {} --normal {}".format(PY3_PATH, ckpy_path, bamfile, stripRefG,
nthreads, outdir, normal)
else:
cmd = "{} {} batch -m wgs -r {} -p {} -d {} {}".format(PY3_PATH, ckpy_path, ckRef, nthreads, outdir, bamfile)
logging.info(cmd + "\n")
call(cmd, shell=True)
metadata_dict["cnvkit_cmd"] = cmd + " ; "
rscript_str = ""
if args.rscript_path:
rscript_str = "--rscript-path " + args.rscript_path
logging.info("Set Rscript flag: " + rscript_str)
cnrFile = outdir + bamBase + ".cnr"
cnsFile = outdir + bamBase + ".cns"
logging.info("Running CNVKit segment")
# TODO: possibly include support for adding VCF calls.
cmd = "{} {} segment {} {} -p {} -m {} -o {}".format(PY3_PATH, ckpy_path, cnrFile, rscript_str, nthreads, seg_meth,
cnsFile)
logging.info(cmd + "\n")
exit_code = call(cmd, shell=True)
if exit_code != 0:
logging.error("CNVKit encountered a non-zero exit status. Exiting...\n")
sys.exit(1)
metadata_dict["cnvkit_cmd"] = metadata_dict["cnvkit_cmd"] + cmd
logging.info("Cleaning up temporary CNVkit files")
cmd = "rm -f {}/*tmp.bed {}/*.cnn {}/*target.bed {}/*.bintest.cns".format(outdir, outdir, outdir, outdir)
logging.info(cmd)
call(cmd, shell=True)
cmd = "gzip -f " + cnrFile
logging.info(cmd)
call(cmd, shell=True)
if normal and not args.ref == "GRCh38_viral":
cmd = "rm " + stripRefG + " " + stripRefG + ".fa"
logging.info(cmd)
call(cmd, shell=True)
# Read the CNVkit .cns files
def convert_cnvkit_cns_to_bed(cnvkit_output_directory, base, cnsfile=None, rescaled=False, nofilter=False):
if cnsfile is None:
if not rescaled:
cnsfile = cnvkit_output_directory + base + ".cns"
else:
cnsfile = cnvkit_output_directory + base + "_rescaled.cns"
with open(cnsfile) as infile, open(cnvkit_output_directory + base + "_CNV_CALLS.bed", 'w') as outfile:
head = next(infile).rstrip().rsplit("\t")
for line in infile:
fields = line.rstrip().rsplit("\t")
# s, e = int(fields[1]), int(fields[2])
cn_r = float(fields[4])
cn = 2 ** (cn_r + 1)
# do not filter on size since amplified_intervals.py will merge small ones.
outline = "\t".join(fields[0:3] + ["CNVkit", str(cn)]) + "\n"
outfile.write(outline)
return cnvkit_output_directory + base + "_CNV_CALLS.bed"
def rescale_cnvkit_calls(ckpy_path, cnvkit_output_directory, base, cnsfile=None, ploidy=None, purity=None):
if purity is None and ploidy is None:
logging.warning("Warning: Rescaling called without --ploidy or --purity. Rescaling will have no effect.")
if cnsfile is None:
cnsfile = cnvkit_output_directory + base + ".cns"
if purity < 0.4:
logging.warning("WARNING! Rescaling a low purity sample may cause many false-positive seed regions!")
cmd = "{} {} call {} -m clonal".format(PY3_PATH, ckpy_path, cnsfile)
if purity:
cmd += " --purity " + str(purity)
if ploidy:
cmd += " --ploidy " + str(ploidy)
cmd += " -o " + cnvkit_output_directory + base + "_rescaled.cns"
logging.info("Rescaling CNVKit calls\n" + cmd)
call(cmd, shell=True)
def run_amplified_intervals(AA_interpreter, CNV_seeds_filename, sorted_bam, output_directory, sname, cngain,
cnsize_min):
logging.info("Running amplified_intervals")
AA_seeds_filename = "{}_AA_CNV_SEEDS".format(output_directory + sname)
cmd = "{} {}/amplified_intervals.py --ref {} --bed {} --bam {} --gain {} --cnsize_min {} --out {}".format(
AA_interpreter, AA_SRC, args.ref, CNV_seeds_filename, sorted_bam, str(cngain), str(cnsize_min),
AA_seeds_filename)
logging.info(cmd + "\n")
exit_code = call(cmd, shell=True)
if exit_code != 0:
logging.error("amplified_intervals.py returned a non-zero exit code. Exiting...\n")
sys.exit(1)
metadata_dict["amplified_intervals_cmd"] = cmd
return AA_seeds_filename + ".bed"
def run_AA(amplified_interval_bed, AA_outdir, sname, args):
AA_interpreter = args.aa_python_interpreter
sorted_bam = args.bam
downsample = args.downsample
ref = args.ref
runmode = args.AA_runmode
extendmode = args.AA_extendmode
insert_sdevs = args.AA_insert_sdevs
sv_vcf = args.sv_vcf
sv_vcf_no_filter = args.sv_vcf_no_filter
AA_version = \
Popen([AA_interpreter, AA_SRC + "/AmpliconArchitect.py", "--version"], stdout=PIPE, stderr=PIPE).communicate()[0].rstrip()
if not AA_version:
AA_version = \
Popen([AA_interpreter, AA_SRC + "/AmpliconArchitect.py", "--version"], stdout=PIPE, stderr=PIPE).communicate()[1].rstrip()
try:
AA_version = AA_version.decode('utf-8')
except UnicodeError:
pass
metadata_dict["AA_version"] = AA_version
cmd = "{} {}/AmpliconArchitect.py --ref {} --downsample {} --bed {} --bam {} --runmode {} --extendmode {} --out {}/{}".format(
AA_interpreter, AA_SRC, ref, str(downsample), amplified_interval_bed, sorted_bam, runmode, extendmode,
AA_outdir, sname)
if insert_sdevs is not None:
cmd += " --insert_sdevs {}".format(str(insert_sdevs))
if sv_vcf:
cmd += " --sv_vcf {}".format(sv_vcf)
if sv_vcf_no_filter:
cmd += " --sv_vcf_no_filter"
logging.info(cmd + "\n")
aa_exit_code = call(cmd, shell=True)
if aa_exit_code != 0:
logging.error("AmpliconArchitect returned a non-zero exit code. Exiting...\n")
sys.exit(1)
metadata_dict["AA_cmd"] = cmd
def run_AC(AA_outdir, sname, ref, AC_outdir, AC_src):
logging.info("Running AC")
# make input file
class_output = AC_outdir + sname
input_file = class_output + ".input"
bed_dir = class_output + "_classification_bed_files/"
if os.path.exists(bed_dir):
logging.warning("WARNING! AC files were not cleared prior to re-running. New classifications may become "
"mixed with previous classification files!")
cmd = "{}/make_input.sh {} {}".format(AC_src, AA_outdir, class_output)
logging.info(cmd)
call(cmd, shell=True)
with open(input_file) as ifile:
sample_info_dict["number_of_AA_amplicons"] = len(ifile.readlines())
cmd = "{} {}/amplicon_classifier.py -i {} --ref {} -o {}".format(PY3_PATH, AC_src, input_file, ref, class_output)
logging.info(cmd + "\n")
call(cmd, shell=True)
metadata_dict["AC_cmd"] = cmd
# Get AC version
AC_version = \
Popen([PY3_PATH, AC_src + "/amplicon_classifier.py", "--version"], stdout=PIPE, stderr=PIPE).communicate()[
0].rstrip()
try:
AC_version = AC_version.decode('utf-8')
except UnicodeError:
pass
metadata_dict["AC_version"] = AC_version
# iterate over the bed files and count anything that isn't "unknown" as a feature
feat_count = 0
if os.path.exists(bed_dir):
for bf in os.listdir(bed_dir):
if not "unknown" in bf and bf.endswith(".bed"):
feat_count += 1
sample_info_dict["number_of_AA_features"] = feat_count
def make_AC_table(sname, AC_outdir, AC_src, run_metadata_file, sample_metadata_file, ref, cnv_bed=None):
# make the AC output table
class_output = AC_outdir + sname
input_file = class_output + ".input"
summary_map_file = class_output + "_summary_map.txt"
classification_file = class_output + "_amplicon_classification_profiles.tsv"
cmd = "{} {}/make_results_table.py -i {} --classification_file {} --summary_map {} --ref {}".format(
PY3_PATH, AC_src, input_file, classification_file, summary_map_file, ref)
if cnv_bed:
cmd += " --cnv_bed " + cnv_bed
if run_metadata_file:
cmd += " --run_metadata_file " + run_metadata_file
if sample_metadata_file:
cmd += " --sample_metadata_file " + sample_metadata_file
logging.info(cmd + "\n")
call(cmd, shell=True)
def get_ref_sizes(ref_genome_size_file):
chr_sizes = {}
with open(ref_genome_size_file) as infile:
for line in infile:
fields = line.rstrip().rsplit()
if fields:
chr_sizes[fields[0]] = str(int(fields[1]) - 1)
return chr_sizes
def get_ref_centromeres(ref_name):
centromere_dict = {}
fnameD = {"GRCh38": "GRCh38_centromere.bed", "GRCh37": "human_g1k_v37_centromere.bed",
"hg19": "hg19_centromere.bed",
"mm10": "mm10_centromere.bed", "GRCm38": "GRCm38_centromere.bed", "GRCh38_viral": "GRCh38_centromere.bed"}
with open(AA_REPO + ref_name + "/" + fnameD[ref_name]) as infile:
for line in infile:
if not "centromere" in line and not "acen" in line:
continue
fields = line.rstrip().rsplit("\t")
if fields[0] not in centromere_dict:
centromere_dict[fields[0]] = (fields[1], fields[2])
else:
pmin = min(int(centromere_dict[fields[0]][0]), int(fields[1]))
pmax = max(int(centromere_dict[fields[0]][1]), int(fields[2]))
# pad with 20kb to avoid freebayes issues in calling near centromeres
centromere_dict[fields[0]] = (str(pmin - 20000), str(pmax + 20000))
return centromere_dict
def save_run_metadata(outdir, sname, args, launchtime, commandstring):
# make a dictionary that stores
# datetime
# hostname
# ref
# PAA command
# AA python interpreter version
# bwa cmd
# CN cmd
# AA cmd
# PAA version
# CNVKit version
# AA version
# AC version
metadata_dict["launch_datetime"] = launchtime
metadata_dict["hostname"] = socket.gethostname()
metadata_dict["ref_genome"] = args.ref
aapint = args.aa_python_interpreter if args.aa_python_interpreter else "python"
aa_python_v = Popen([aapint, "--version"], stdout=PIPE, stderr=PIPE).communicate()[0].rstrip()
try:
aa_python_v = aa_python_v.decode('utf-8')
except UnicodeError:
pass
metadata_dict["AA_python_version"] = aa_python_v
metadata_dict["AmpliconSuite-pipeline_command"] = commandstring
metadata_dict["AmpliconSuite-pipeline_version"] = __ampliconsuitepipeline_version__
metadata_dict["Samtools version"] = "{}.{}".format(samtools_version[0], samtools_version[1])
for x in ["bwa_cmd", "cnvkit_cmd", "amplified_intervals_cmd", "AA_cmd", "AC_cmd", "cnvkit_version", "AA_version",
"AC_version"]:
if x not in metadata_dict:
metadata_dict[x] = "NA"
# save the json dict
run_metadata_filename = outdir + sname + "_run_metadata.json"
with open(run_metadata_filename, 'w') as fp:
json.dump(metadata_dict, fp, indent=2)
# sample_info_dict["run_metadata_file"] = run_metadata_filename
return run_metadata_filename
def detect_run_failure(align_stderr_file, AA_outdir, sname, AC_outdir):
if align_stderr_file:
cmd = 'grep -i error ' + align_stderr_file
try:
aln_errs = check_output(cmd, shell=True).decode("utf-8")
except CalledProcessError:
aln_errs = ""
if aln_errs:
logging.error("Detected error during bwa mem alignment stage\n")
return True
if AA_outdir:
sumfile = AA_outdir + sname + "_summary.txt"
if os.path.isfile(sumfile):
namps = -1
with open(sumfile) as infile:
for line in infile:
if line.startswith("#Amplicons = "):
namps = int(line.rstrip().rsplit(" = ")[-1])
break
if namps < 0:
logging.error("Detected truncated or missing AA outputs")
return True
for x in range(1, namps + 1):
try:
fsize = os.stat(AA_outdir + sname + "_amplicon" + str(x) + "_cycles.txt").st_size
except OSError:
fsize = 0
if fsize == 0:
logging.error("Detected truncated or missing AA outputs")
return True
else:
logging.error("Detected error during AA stage")
return True
if AC_outdir:
try:
fsize1 = os.stat(AC_outdir + sname + "_amplicon_classification_profiles.tsv").st_size
fsize2 = os.stat(AC_outdir + sname + "_result_table.tsv").st_size
except OSError:
fsize1 = 0
fsize2 = 0
if fsize1 == 0 or fsize2 == 0:
logging.error("Detected error during AC stage\n")
return True
return False
def get_samtools_version(samtools):
try:
# Run the command to get the version information
result = Popen([samtools], stderr=PIPE, stdout=PIPE)
_, output = result.communicate()
# Decode the output if it's in bytes (Python 3)
if isinstance(output, bytes):
output = output.decode('utf-8')
# Parse the version information to extract major and minor versions
version_pattern = r'Version: (\d+)\.(\d+)'
match = re.search(version_pattern, output)
if match:
major_version = int(match.group(1))
minor_version = int(match.group(2))
return major_version, minor_version
else:
# Return None if version information couldn't be parsed
return None, None
except OSError as e:
# Handle the case when Samtools is not found
logging.error("Error: Samtools not found. Please make sure it is installed and in your PATH.")
return None, None
def download_file(url, destination_folder):
import urllib.request # here because python2 not work with it
filename = os.path.join(destination_folder, url.split("/")[-1])
try:
response = urllib.request.urlopen(url)
file_size = int(response.headers.get('Content-Length', 0))
response.close()
file_size = round(file_size / (1024**3), 2)
if file_size > 0.1:
print("\nDownloading " + url + " ... (" + str(file_size) + "GB)")
else:
print("\nDownloading " + url + " ...")
urllib.request.urlretrieve(url, filename)
print("File downloaded and saved to: " + str(filename))
except Exception as e:
print("Failed to download file. Error: " + str(e))
def extract_tar_gz(file_path, destination_folder):
if not file_path.endswith('.tar.gz'):
sys.stderr.write("Cannot extract file " + file_path)
sys.exit(1)
with tarfile.open(file_path, 'r:gz') as tar:
tar.extractall(destination_folder)
os.remove(file_path)
def contains_spaces(file_path):
return any(char == ' ' for char in file_path)
# MAIN #
if __name__ == '__main__':
# Parses the command line arguments
parser = argparse.ArgumentParser(
description="A pipeline wrapper for AmpliconArchitect, invoking alignment CNV calling and CNV filtering prior. "
"Can launch AA, as well as downstream amplicon classification.")
parser.add_argument("-v", "--version", action='version',
version='AmpliconSuite-pipeline version {version} \n'.format(version=__ampliconsuitepipeline_version__))
parser.add_argument("--download_repo", help="Download the selected data repo to the $AA_DATA_REPO "
"directory and exit. '_indexed' suffix indicates BWA index is included, which is useful if "
"performing alignment with AmpliconSuite-pipeline, but has a larger filesize.", choices=["hg19",
"GRCh37", "GRCh38", "mm10", "GRCh38_viral", "hg19_indexed", "GRCh37_indexed", "GRCh38_indexed",
"mm10_indexed", "GRCh38_viral_indexed"], nargs='+')
parser.add_argument("-o", "--output_directory", metavar='PATH', help="output directory names (will create if not already created)")
parser.add_argument("-s", "--sample_name", metavar='STR', help="(Required) Sample name")
parser.add_argument("-t", "--nthreads", metavar='INT', help="(Required) Number of threads to use in BWA and CNV calling")
parser.add_argument("--run_AA", help="Run AA after all files prepared. Default off.", action='store_true')
parser.add_argument("--run_AC", help="Run AmpliconClassifier after all files prepared. Default off.",
action='store_true')
parser.add_argument("--ref", metavar='STR', help="Reference genome version.", choices=["hg19", "GRCh37",
"GRCh38", "hg38", "mm10", "GRCm38", "GRCh38_viral"])
parser.add_argument("--cngain", metavar='FLOAT', type=float, help="CN gain threshold to consider for AA seeding",
default=4.5)
parser.add_argument("--cnsize_min", metavar='INT', type=int, help="CN interval size (in bp) to consider for AA seeding",
default=50000)
parser.add_argument("--downsample", metavar='FLOAT', type=float, help="AA downsample argument (see AA documentation)",
default=10)
parser.add_argument("--rscript_path", metavar='PATH', help="Specify custom path to Rscript for CNVKit, "
"which requires R version >=3.5")
parser.add_argument("--python3_path", metavar='PATH', help="If needed, specify a custom path to python3.")
parser.add_argument("--aa_python_interpreter",
help="By default AmpliconSuite-pipeline will use the system's default python path. If you would like to use "
"a different python version with AA, set this to either the path to the interpreter or "
"'python3' or 'python2' (default 'python')", metavar='PATH', type=str, default='python')
parser.add_argument("--sv_vcf",
help="Provide a VCF file of externally-called SVs to augment SVs identified by AA internally.",
metavar='FILE', action='store', type=str)
parser.add_argument("--sv_vcf_no_filter", help="Use all external SV calls from the --sv_vcf arg, even "
"those without 'PASS' in the FILTER column.", action='store_true', default=False)
parser.add_argument("--AA_src", metavar='PATH', help="Specify a custom $AA_SRC path. Overrides the bash variable")
parser.add_argument("--AA_runmode", metavar='STR', help="If --run_AA selected, set the --runmode argument to AA. Default mode is "
"'FULL'", choices=['FULL', 'BPGRAPH', 'CYCLES', 'SVVIEW'], default='FULL')
parser.add_argument("--AA_extendmode", metavar='STR', help="If --run_AA selected, set the --extendmode argument to AA. Default "
"mode is 'EXPLORE'", choices=["EXPLORE", "CLUSTERED", "UNCLUSTERED", "VIRAL"], default='EXPLORE')
parser.add_argument("--AA_insert_sdevs", help="Number of standard deviations around the insert size. May need to "
"increase for sequencing runs with high variance after insert size selection step. (default "
"3.0)", metavar="FLOAT", type=float, default=None)
parser.add_argument("--normal_bam", metavar='FILE', help="Path to matched normal bam for CNVKit (optional)")
parser.add_argument("--ploidy", metavar='FLOAT', type=float, help="Ploidy estimate for CNVKit (optional). This is not used outside of CNVKit.",
default=None)
parser.add_argument("--purity", metavar='FLOAT', type=float, help="Tumor purity estimate for CNVKit (optional). This is not used outside of CNVKit.",
default=None)
parser.add_argument("--cnvkit_segmentation", metavar='STR', help="Segmentation method for CNVKit (if used), defaults to CNVKit "
"default segmentation method (cbs).", choices=['cbs', 'haar', 'hmm', 'hmm-tumor', 'hmm-germline', 'none'],
default='cbs')
parser.add_argument("--no_filter", help="Do not run amplified_intervals.py to remove low confidence candidate seed"
" regions overlapping repetitive parts of the genome", action='store_true')
parser.add_argument("--no_QC", help="Skip QC on the BAM file. Do not adjust AA insert_sdevs for "
"poor-quality insert size distribution", action='store_true')
parser.add_argument("--sample_metadata", metavar='FILE', help="JSON file of sample metadata to build on")
parser.add_argument("--samtools_path", help="Path to samtools binary (e.g., /path/to/my/samtools). If unset, will use samtools on system path.",
default='')
group = parser.add_mutually_exclusive_group()
group.add_argument("--bam", "--sorted_bam", metavar='FILE', help="Coordinate sorted BAM file (aligned to an AA-supported "
"reference.)")
group.add_argument("--fastqs", metavar='TWO FILES', help="Fastq files (r1.fq r2.fq)", nargs=2)
group.add_argument("--completed_AA_runs", metavar='PATH',
help="Path to a directory containing one or more completed AA runs which utilized the same reference genome.")
group2 = parser.add_mutually_exclusive_group()
group2.add_argument("--cnv_bed", "--bed", metavar='FILE',
help="BED file (or CNVKit .cns file) of CNV changes. Fields in the bed file should"
" be: chr start end name cngain")
group2.add_argument("--cnvkit_dir", metavar='PATH', help="Path to cnvkit.py. Assumes CNVKit is on the system path if not set",
default="")
group2.add_argument("--completed_run_metadata", metavar='FILE',
help="Run metadata JSON to retroactively assign to collection of samples", default="")
group2.add_argument("--align_only", help="Only perform the alignment stage (do not run CNV calling and seeding",
action='store_true')
# start timing
ta = time.time()
ti = ta
launchtime = str(datetime.now())
args = parser.parse_args()
# Check if AA_REPO set, print error and quit if not
try:
AA_REPO = os.environ['AA_DATA_REPO'] + "/"
except KeyError:
sys.stderr.write("AA_DATA_REPO bash variable not found. Please see installation instructions and run ./install.sh before using.\n")
sys.exit(1)
# Download any requested data repo files
if args.download_repo:
# launch data repo download and exit
data_repo_base_url = "https://datasets.genepattern.org/data/module_support_files/AmpliconArchitect/"
for ref in args.download_repo:
print(ref)
ref_base_url = data_repo_base_url + ref
md5file = ref_base_url + "_md5sum.txt"
ref_file = ref_base_url + ".tar.gz"
if os.path.exists(AA_REPO + ref):
print("An AA data repo directory already exists for " + ref + " and it will be replaced!")
download_file(md5file, AA_REPO)
download_file(ref_file, AA_REPO)
print("Extracting...\n")
extract_tar_gz(AA_REPO + ref + ".tar.gz", AA_REPO)
print("Finished")
sys.exit(0)
# Preflight checks for running AS-pipeline
if not args.sample_name:
parser.error("--sample_name (-s) is a required argument.")
if not args.nthreads:
parser.error("--nthreads (-t) is a required argument.")
if not any([args.bam, args.fastqs, args.completed_AA_runs]):
parser.error("One of --bam | --fastqs | --completed_AA_runs is required!")
# set an output directory if user did not specify
if not args.output_directory:
args.output_directory = os.getcwd()
if not args.output_directory.endswith("/"):
args.output_directory += "/"
sname = args.sample_name
outdir = args.output_directory
sample_metadata_filename = args.output_directory + sname + "_sample_metadata.json"
# set samtools version for use
if not args.samtools_path.endswith("/samtools"):
if args.samtools_path and not args.samtools_path.endswith("/"):
args.samtools_path += "/"
args.samtools_path += "samtools"
# Make and clear necessary directories.
# make the output directory location if it does not exist
if not os.path.exists(args.output_directory):
os.mkdir(args.output_directory)
# initiate logging
paa_logfile = args.output_directory + sname + '.log'
logging.basicConfig(filename=paa_logfile, format='[%(name)s:%(levelname)s]\t%(message)s',
level=logging.INFO, filemode='w')
console_handler = logging.StreamHandler()
console_handler.setLevel(logging.INFO)
formatter = logging.Formatter('[%(name)s:%(levelname)s]\t%(message)s')
console_handler.setFormatter(formatter)
logging.getLogger().addHandler(console_handler)
logging.info("Launched on " + launchtime)
logging.info("AmpiconSuite-pipeline version " + __ampliconsuitepipeline_version__ + "\n")
commandstring = ""
for arg in sys.argv:
if ' ' in arg:
commandstring += '"{}" '.format(arg)
else:
commandstring += "{} ".format(arg)
logging.info("AmpliconSuite-pipeline command:")
logging.info(commandstring + "\n")
if "/" in args.sample_name:
logging.error("Sample name -s cannot be a path. Specify output directory with -o.\n")
sys.exit(1)
finish_flag_filename = args.output_directory + args.sample_name + "_finish_flag.txt"
if os.path.exists(finish_flag_filename):
logging.warning("WARNING: Running AmpliconSuite-pipeline.py with outputs directed into the same output location"
" as a previous run may cause crashes or other unexpected behavior. To avoid errors, clear "
"previous files before re-running.\n")
with open(finish_flag_filename, 'w') as ffof:
ffof.write("UNSUCCESSFUL\n")
timing_logfile = open(args.output_directory + args.sample_name + '_timing_log.txt', 'w')
timing_logfile.write("#stage:\twalltime(seconds)\n")
samtools_version = get_samtools_version(args.samtools_path)
if samtools_version:
logging.debug("Samtools version: {}.{}".format(samtools_version[0], samtools_version[1]))
else:
logging.error("Failed to retrieve Samtools version.")
sys.exit(1)
# Check if expected system paths and files are present. Check if provided argument combinations are valid.
if args.AA_src:
os.environ['AA_SRC'] = args.AA_src
if not os.path.exists(os.path.join(AA_REPO, "coverage.stats")):
logging.info("coverage.stats file not found in " + AA_REPO + "\nCreating a new coverage.stats file.")
cmd = "touch {}coverage.stats && chmod a+rw {}coverage.stats".format(AA_REPO, AA_REPO)
logging.info(cmd)
call(cmd, shell=True)
try:
AA_SRC = os.environ['AA_SRC']
except KeyError:
try:
import ampliconarchitectlib
AA_SRC = os.path.realpath(os.path.dirname(ampliconarchitectlib.__file__))
except ModuleNotFoundError:
logging.error("AA_SRC bash variable or library files not found. AmpliconArchitect may not be properly installed.\n")
sys.exit(1)
try:
AC_SRC = os.environ['AC_SRC']
except KeyError:
try:
import ampclasslib
ac_path = check_output("which amplicon_classifier.py", shell=True).decode("utf-8")
AC_SRC = ac_path.rsplit("/amplicon_classifier.py")[0]
except Exception as e:
logging.error(e)
logging.error(
"\nAC_SRC bash variable or library files not found. AmpliconClassifier may not be properly installed.\n")
sys.exit(1)
if (args.fastqs or args.completed_AA_runs) and not args.ref:
logging.error("Must specify --ref when providing unaligned fastq files or completed AA runs.\n")
sys.exit(1)
if args.completed_run_metadata.lower() == "none":
args.completed_run_metadata = None
# if not these args are set, assume cnvkit.py is on the path.
if not (args.cnv_bed or args.cnvkit_dir or args.completed_run_metadata or args.align_only) and (args.fastqs or
args.bam):
try:
args.cnvkit_dir = str(check_output(["which cnvkit.py"], shell=True).decode("utf-8").rstrip())
except CalledProcessError:
logging.error("cnvkit.py not found on system path. Must specify --cnvkit_dir")
sys.exit(1)
elif args.cnvkit_dir and not args.cnvkit_dir.endswith("/") and not args.cnvkit_dir.endswith("cnvkit.py"):
args.cnvkit_dir += "/"
else:
args.completed_run_metadata = None
if not args.cnvkit_dir.endswith("cnvkit.py"):
args.cnvkit_dir += "cnvkit.py"
if args.run_AA:
if not os.path.exists(os.environ["HOME"] + "/mosek/mosek.lic") and not "MOSEKLM_LICENSE_FILE" in os.environ:
logging.error("--run_AA set, but MOSEK license not found in $HOME/mosek/")
sys.exit(1)
elif "MOSEKLM_LICENSE_FILE" in os.environ:
if os.environ["MOSEKLM_LICENSE_FILE"].endswith("mosek.lic"):
logging.error("MOSEKLM_LICENSE_FILE should be the path of the directory of the license, not the full path. Please update your .bashrc, and run 'source ~/.bashrc'")
sys.exit(1)
elif not os.path.exists(os.environ["MOSEKLM_LICENSE_FILE"] + "/mosek.lic"):
logging.error("--run_AA set, but MOSEK license not found in " + os.environ["MOSEKLM_LICENSE_FILE"])
sys.exit(1)
runCNV = None
if args.cnvkit_dir and not args.cnv_bed:
runCNV = "CNVkit"
# check Rscript version
test_rscript = "Rscript"
if args.rscript_path:
if not args.rscript_path.endswith("/Rscript"):
args.rscript_path += "/Rscript"
test_rscript = args.rscript_path
try:
rscript_version_out = str(check_output([test_rscript, "--version"], stderr=STDOUT).decode("utf-8").rstrip())
except CalledProcessError:
logging.error(test_rscript + " not found. Must specify --rscript_path")
sys.exit(1)
if args.python3_path:
if not args.python3_path.endswith("/python") and not args.python3_path.endswith("/python3"):
args.python3_path += "/python3"
PY3_PATH = args.python3_path
if args.aa_python_interpreter and not any(args.aa_python_interpreter.endswith(x) for x in ['python', 'python2', 'python3']):
logging.error("--aa_python_interpreter must be a path of a valid python interpreter")
sys.exit(1)
refFnames = {x: None for x in ["hg19", "GRCh37", "GRCh38", "GRCh38_viral", "mm10"]}
# Paths of all the repo files needed
if args.ref == "hg38":
args.ref = "GRCh38"
if args.ref == "GRCm38":
args.ref = "mm10"
for rname in refFnames.keys():
if os.path.exists(AA_REPO + "/" + rname):
refFnames[rname] = check_reference.get_ref_fname(AA_REPO, rname)
faidict = {}
if args.bam:
if contains_spaces(args.bam):
logging.error("BAM filepath cannot contain spaces!")
sys.exit(1)
if args.ref and refFnames[args.ref]:
faidict[args.ref] = AA_REPO + args.ref + "/" + refFnames[args.ref] + ".fai"
elif args.ref and refFnames[args.ref] is None:
em = "Data repo files for ref " + args.ref + " not found. Please download using the '--download_repo " + args.ref + "' option\n"
logging.error(em)
sys.exit(1)
else:
for k, v in refFnames.items():
if v:
faidict[k] = AA_REPO + k + "/" + v + ".fai"
determined_ref = check_reference.check_ref(args.bam, faidict, args.samtools_path)
if not determined_ref and not args.ref:
logging.error("Could not determine ref build. Please make sure AA data repo is populated.")
sys.exit(1)
elif not args.ref:
args.ref = determined_ref
elif args.ref and not determined_ref:
logging.warning("WARNING! The BAM file did not match " + args.ref)
try:
with open(AA_REPO + args.ref + "/last_updated.txt", 'r') as file:
datestring = file.read()
logging.info(args.ref + " data repo constructed on " + datestring)
except FileNotFoundError:
logging.warning("Data repo appears to be out of date. Please update your data repo!\n")
gdir = AA_REPO + args.ref + "/"
ref_fasta = gdir + refFnames[args.ref]
ref_genome_size_file = gdir + args.ref + "_noAlt.fa.fai"
if args.cnv_bed and not os.path.isfile(args.cnv_bed):
logging.error("Specified CNV bed file does not exist: " + args.cnv_bed + "\n")
sys.exit(1)
if not args.sample_metadata:
args.sample_metadata = os.path.realpath(os.path.dirname(check_reference.__file__)) + "/sample_metadata_skeleton.json"
with open(args.sample_metadata) as input_json:
sample_info_dict = json.load(input_json)
sample_info_dict["reference_genome"] = args.ref
sample_info_dict["sample_name"] = sname
tb = time.time()
timing_logfile.write("Initialization:\t" + "{:.2f}".format(tb - ta) + "\n")
ta = tb
logging.info("Running AmpliconSuite-pipeline on sample: " + sname)
# Begin pipeline
aln_stage_stderr = None
if args.fastqs:
# Run BWA
if args.fastqs[0] == args.fastqs[1]:
logging.error(str(args.fastqs))
logging.error("You must provide two different fastq files for paired-end reads!\n")
sys.exit(1)
elif contains_spaces(args.fastqs[0]) or contains_spaces(args.fastqs[1]):
logging.error("FASTQ filepaths cannot contain spaces!")
sys.exit(1)
fastqs = " ".join(args.fastqs)
logging.info("Will perform alignment on " + fastqs)
args.bam, aln_stage_stderr = run_bwa(ref_fasta, fastqs, outdir, sname, args.nthreads, args.samtools_path, samtools_version)
if not args.completed_AA_runs:
bamBaiNoExt = args.bam[:-3] + "bai"
cramCraiNoExt = args.bam[:-4] + "crai"
baiExists = os.path.isfile(args.bam + ".bai") or os.path.isfile(bamBaiNoExt)
craiExists = os.path.isfile(args.bam + ".crai") or os.path.isfile(cramCraiNoExt)
if not baiExists and not craiExists:
logging.info(args.bam + " index not found, calling samtools index")
call([args.samtools_path, "index", args.bam])
logging.info("Finished indexing")
bambase = os.path.splitext(os.path.basename(args.bam))[0]
prop_paired_proportion = None
if not args.no_QC:
logging.debug("samtools path is set to: " + args.samtools_path)
prop_paired_proportion = check_reference.check_properly_paired(args.bam, args.samtools_path)
tb = time.time()
timing_logfile.write("Alignment, indexing and QC:\t" + "{:.2f}".format(tb - ta) + "\n")
if args.align_only:
logging.info("Completed\n")
tf = time.time()
timing_logfile.write("Total_elapsed_walltime\t" + "{:.2f}".format(tf - ti) + "\n")
timing_logfile.close()
sys.exit()
ta = tb
centromere_dict = get_ref_centromeres(args.ref)
chr_sizes = get_ref_sizes(ref_genome_size_file)
# coordinate CNV calling
cnvkit_output_directory = None
if runCNV == "CNVkit":
cnvkit_output_directory = args.output_directory + sname + "_cnvkit_output/"
if not os.path.exists(cnvkit_output_directory):
os.mkdir(cnvkit_output_directory)
run_cnvkit(args.cnvkit_dir, args.nthreads, cnvkit_output_directory, args.bam,
seg_meth=args.cnvkit_segmentation, normal=args.normal_bam, ref_fasta=ref_fasta)
if args.ploidy or args.purity:
rescale_cnvkit_calls(args.cnvkit_dir, cnvkit_output_directory, bambase, ploidy=args.ploidy,
purity=args.purity)
rescaling = True
else:
rescaling = False
args.cnv_bed = convert_cnvkit_cns_to_bed(cnvkit_output_directory, bambase, rescaled=rescaling)
if args.cnv_bed.endswith(".cns"):