Merge pull request #58 from AntonJMLarsson/master

Speed up bam splitting for bcall
caleblareau · Jun 1, 2022 · 9606660 · 9606660
2 parents f0a27c1 b0c1650
commit 9606660
Showing 1 changed file with 12 additions and 11 deletions.
diff --git a/mgatk/bin/python/split_barcoded_bam.py b/mgatk/bin/python/split_barcoded_bam.py
@@ -16,35  16,35 @@
 base=os.path.basename(bamfile)
 basename=os.path.splitext(base)[0]
 
-def getBarcode(intags):
 def getBarcode(read):
 	'''
 	Parse out the barcode per-read
 	'''
-	for tg in intags:
-		if(barcodeTag == tg[0]):
-			return(tg[1])
-	return("NA")
 	if read.has_tag(barcodeTag):
 		return read.get_tag(barcodeTag)
 	else:
 		return "NA"
 
 
-def writePassingReads(bc, mtchr):
 def writePassingReads(bc_dict, mtchr):
 	'''
 	Write out reads to their corresponding files based on a barcode index
 	'''
 	bam = pysam.AlignmentFile(bamfile,'rb')
 	Itr = bam.fetch(str(mtchr),multiple_iterators=False)
 	for read in Itr:
-		read_barcode = getBarcode(read.tags)
 		read_barcode = getBarcode(read)
 
 		# If read barcode is in whitelist, then write it out
-		if read_barcode in bc:
-			idx = bc.index(read_barcode)
 		if read_barcode in bc_dict:
 			idx = bc_dict[read_barcode]
 			file = fopen[idx]
 			file.write(read)
 
 # Read in the barcodes
 with open(bcfile) as barcode_file_handle:
     content = barcode_file_handle.readlines()
-bc = [x.strip() for x in content] 
 bc = [x.strip() for x in content]
 
 # Open up a bunch of files and write out reads for valid barcodes
 @contextmanager
@@ -61,8  61,9 @@ def multi_file_manager(files, mode='rt'):
 
 # Final loop to write out passing reads
 bambcfiles = [outfolder   "/"   bc1   ".bam" for bc1 in bc]
 bc_dict = {bc1: i  for i,bc1 in enumerate(bc)}
 with multi_file_manager(bambcfiles) as fopen:
-	writePassingReads(bc, mtchr)
 	writePassingReads(bc_dict, mtchr)