def main():
# Parse command line arguments
args = parse_args()
bamfile = args.bam
bedfile = args.bed
outfile = args.output
if outfile:
outstream = open(outfile, 'w')
else:
outstream = sys.stdout
max_distance = args.dist
out_header = '\t'.join(
['STR_chr', 'STR_start', 'STR_stop', 'motif', 'reflen', 'count'])
outstream.write(out_header + '\n')
#STR_bed = parse_bed(args.bed, position_base=0)
STR_bed = bt.BedTool(bedfile)
readlen = detect_readlen(bamfile)
# Read bam
bam = pysam.Samfile(bamfile, 'rb')
# Get relevant chromosomes
required_chroms = []
unpaired = 0
total = 0
for chrom in bam.references:
if chrom.startswith('STR-'):
required_chroms.append(chrom)
for chrom in required_chroms:
motif = chrom.split('-')[1]
all_positions = []
all_segments = bam.fetch(reference=chrom)
for read in all_segments:
total += 1
try:
mate_chr = read.next_reference_name
except ValueError:
unpaired += 1
continue
mate_start = read.next_reference_start
mate_stop = mate_start + readlen
all_positions.append([mate_chr, mate_start, mate_stop])
# Strategy:
# Merge all overlapping intervals
# Keep the count of reads corresponding to each merged interval (i.e. 1 for each read contained in it)
# Assign each interval to the closest STR (within 500 bp? - the insert size) with the correct motif, adding together the count of reads
# Check motif: == normalise_str()
# There should be two 1-2 intervals per STR, likely one for each flank.
# Report the read count for each STR
if len(all_positions) > 0:
motif_bed = bt.BedTool(all_positions).sort()
# Merge all the intervals, then count how many of the original intervals overlap the merged ones (4th column)
motif_coverage = motif_bed.merge(stream=True).coverage(b=motif_bed,
counts=True)
tmp_bed = 'tmp-' + randomletters(
8
) + '.bed' #create temporary file for bedtools to write to and pandas to read since streams don't seem to work
closest_STR = motif_coverage.closest(STR_bed, d=True,
stream=True).saveas(tmp_bed)
colnames = [
'chr', 'start', 'stop', 'count', 'STR_chr', 'STR_start',
'STR_stop', 'motif', 'reflen', 'distance'
]
df = pd.read_csv(tmp_bed, sep='\t', header=None, names=colnames)
os.remove(tmp_bed) #delete temporary file
# Filter out loci that are too far away
df = df.loc[df['distance'] <= max_distance, :]
df['motif'] = df['motif'].map(
normalise_str) # Normalise the STR motif to enable comparisons
# Remove STRs that don't match the motif
df = df.loc[df['motif'] == normalise_str(motif), :]
df = df.loc[:, [
'STR_chr', 'STR_start', 'STR_stop', 'motif', 'count', 'reflen'
]]
summed = df.groupby(
['STR_chr', 'STR_start', 'STR_stop', 'motif', 'reflen'],
as_index=False).aggregate(np.sum)
summed.to_csv(outstream, sep='\t', header=False, index=False)
outstream.close()
if unpaired == total:
sys.exit(
'ERROR: all reads tested appear to be unpaired. You may wish to check your bam file is paired end and correctly formed.'
)
elif unpaired > 0:
sys.stderr.write(
'WARNING: it appears that {} of {} reads checked were unpaired and so no useful data could be obtained from them.\n'
)
def locus_counts(bamfiles, bedfile, outfile, max_distance):
# Check bamfiles have unique names
#print(bamfiles, type(bamfiles))
if not isinstance(bamfiles, list):
raise TypeError('Expecting a list, got {}'.format(type(bamfiles)))
if len(set(bamfiles)) < len(bamfiles):
sys.exit(
'ERROR: There were multiple bamfiles with the same filename. Please check your input'
)
#STR_bed = parse_bed(args.bed, position_base=0)
STR_bed = bt.BedTool(bedfile).sort()
all_results = []
for bamfile in bamfiles:
readlen, count_noCIGAR = detect_readlen(bamfile)
# Read bam
bam = pysam.Samfile(bamfile, 'rb')
# Get relevant chromosomes
required_chroms = []
unpaired = 0
total = 0
for chrom in bam.references:
if chrom.startswith('STR-'):
required_chroms.append(chrom)
# Check if any STR- chromosomes
if len(required_chroms) == 0:
sys.exit(
'ERROR: There were no reads mapping to chromosomes with names starting with "STR-" in {0}. Are you sure this data is mapped to a reference genome with STR decoy chromosomes?'
.format(bamfile))
for chrom in required_chroms:
motif = chrom.split('-')[1]
all_positions = []
all_segments = bam.fetch(reference=chrom)
for read in all_segments:
#if read.is_secondary:
# continue
total += 1
try:
mate_chr = read.next_reference_name
except ValueError:
unpaired += 1
continue
mate_start = read.next_reference_start
mate_stop = mate_start + readlen
all_positions.append([mate_chr, mate_start, mate_stop])
# Strategy:
# Merge all overlapping intervals
# Keep the count of reads corresponding to each merged interval (i.e. 1 for each read contained in it)
# Assign each interval to the closest STR (within 500 bp? - the insert size) with the correct motif, adding together the count of reads
# Check motif: == normalise_str()
# There should be two 1-2 intervals per STR, likely one for each flank.
# Report the read count for each STR
if len(all_positions) > 0:
motif_bed = bt.BedTool(all_positions).sort()
# Merge all the intervals, then count how many of the original intervals overlap the merged ones (4th column)
motif_coverage = motif_bed.merge(stream=True).coverage(
b=motif_bed, counts=True, nonamecheck=True)
tmp_bed = 'tmp-' + randomletters(
8
) + '.bed' #create temporary file for bedtools to write to and pandas to read since streams don't seem to work
closest_STR = motif_coverage.closest(
STR_bed, d=True, stream=True,
nonamecheck=True).saveas(tmp_bed)
colnames = [
'chr', 'start', 'stop', 'count', 'STR_chr', 'STR_start',
'STR_stop', 'motif', 'reflen', 'distance'
]
df = pd.read_csv(tmp_bed,
sep='\t',
header=None,
names=colnames)
os.remove(tmp_bed) #delete temporary file
# Filter out loci that are too far away
df = df.loc[df['distance'] <= max_distance, :]
df['motif'] = df['motif'].map(
normalise_str
) # Normalise the STR motif to enable comparisons
# Remove STRs that don't match the motif
df = df.loc[df['motif'] == normalise_str(motif), :]
df = df.loc[:, [
'STR_chr', 'STR_start', 'STR_stop', 'motif', 'count',
'reflen'
]]
all_results.append(df)
if total == 0:
sys.exit(
'ERROR: there were no reads overlapping the target STR regions. This may indicate a problem with the input file.\n'
)
elif unpaired == total:
sys.exit(
'ERROR: all {0} reads overlapping the target STR regions appear to be unpaired. You may wish to check your bam file is paired-end and correctly formed.\n'
.format(total))
elif unpaired > 0:
sys.stderr.write(
'WARNING: it appears that {0} of the {1} reads overlapping the target STR regions were unpaired and so no useful data could be obtained from them.\n'
.format(unpaired, total))
# Sum counts from multiple bam files and multiple rows
if len(all_results) == 1:
df_total = all_results[0]
else:
df_total = pd.concat(all_results, ignore_index=True)
summed = df_total.groupby(
['STR_chr', 'STR_start', 'STR_stop', 'motif', 'reflen'],
as_index=False).aggregate(np.sum)
# Print a warning message in case of reads without a CIGAR string
if count_noCIGAR > 0:
sys.stderr.write('WARNING: ' + str(count_noCIGAR) + ' read(s) in ' +
bamfile + ' file had no CIGAR string.\n')
if total == 0:
sys.exit(
'ERROR: there were no reads overlapping the target STR regions. This may indicate a problem with the input file.\n'
)
elif unpaired == total:
sys.exit(
'ERROR: all {0} reads overlapping the target STR regions appear to be unpaired. You may wish to check your bam file is paired-end and correctly formed.\n'
.format(total))
elif unpaired > 0:
sys.stderr.write(
'WARNING: it appears that {0} of the {1} reads overlapping the target STR regions were unpaired and so no useful data could be obtained from them.\n'
.format(unpaired, total))
# Write results
if outfile:
outstream = open(outfile, 'w')
else:
outstream = sys.stdout
out_header = '\t'.join(
['STR_chr', 'STR_start', 'STR_stop', 'motif', 'reflen', 'count'])
outstream.write(out_header + '\n')
outstring = summed.to_csv(sep='\t', header=False, index=False)
outstream.write(outstring)
outstream.close()