def get_freqs(path, contig, pos, ref, alt, tlen_cutoff, fm_cutoff):
it = pysam.AlignmentFile(path).pileup(contig, pos)
column = it.next()
while column.pos < pos:
column = it.next()
ref_counts = freq(get_tlens(column, ref, tlen_cutoff, fm_cutoff))
alt_counts = freq(get_tlens(column, alt, tlen_cutoff, fm_cutoff))
all_counts = freq(get_tlens(column, "ALL", tlen_cutoff, fm_cutoff))
return {"ref": ref_counts, "alt": alt_counts, "all": all_counts}
def test_freq(self):
self.assertEqual(freq("ACGTTTAAA"), {
'A': 4,
'C': 1,
'G': 1,
'T': 3
})
def fill_tables(self):
self.nucs = set(pr.alignment.seq[pr.query_position] for
pr in self.pc.pileups)
self.tables = {nuc: freq(pr.alignment.opt("FM") for pr in
self.pc.pileups if
pr.alignment.seq[pr.query_position] == nuc)
for nuc in self.nucs}
def __init__(self, pileup_column, minFM=2, ref='C', alt='T'):
all_counts = freq(i.alignment.seq[i.query_position]
for i in pileup_column.pileups)
fm_counts = freq(i.alignment.seq[i.query_position]
for i in pileup_column.pileups
if i.alignment.opt("FM") >= minFM)
if alt not in all_counts:
all_counts[alt] = 0
if alt not in fm_counts:
fm_counts[alt] = 0
self.all = AAFObj(all_counts[alt], sum(all_counts.values()))
self.fm = AAFObj(fm_counts[alt], sum(fm_counts.values()))
self.all_discounted = AAFObj(fm_counts[alt], sum(all_counts.values()))
bc_stacks = get_bc_stack(pileup_column)
self.all_expected = AAFObj(
get_expectation_count(bc_stacks, alt, minFM),
sum(all_counts.values()))
def fill_tables(self):
self.nucs = set(pr.alignment.seq[pr.query_position]
for pr in self.pc.pileups)
self.tables = {
nuc: freq(
pr.alignment.opt("FM") for pr in self.pc.pileups
if pr.alignment.seq[pr.query_position] == nuc)
for nuc in self.nucs
}
def aaf_pos(bam, contig, pos, ref='C', alt='T', minFM=1):
a = pysam.AlignmentFile(bam)
b = a.pileup(contig, pos)
c = b.next()
while c.pos < pos:
c = b.next()
counts = freq(i.alignment.seq[i.query_position] for i in c.pileups
if i.alignment.opt("FM") >= minFM)
try:
return float(counts[alt]) / counts[ref]
except KeyError:
if ref not in counts:
return -137.
return 0.
def aaf_ret(bam, contig, pos, ref='C', alt='T', minFM=1):
"""
Returns the frequency of ref allele 'C' in pileup in a bam.
"""
a = pysam.AlignmentFile(bam)
b = a.pileup(contig, pos)
c = b.next()
while c.pos < pos:
c = b.next()
counts = freq(i.alignment.seq[i.query_position] for i in c.pileups
if i.alignment.opt("FM") >= minFM)
try:
return counts[alt] / counts[ref]
except KeyError:
if ref not in counts:
return -137.
return 0.