def call_variants(self, ksize, mindist=6, logstream=sys.stderr):
"""Attempt to call variants from this contig alignment.
If the alignment CIGAR matches a known pattern, the appropriate caller
is invoked (SNV or INDEL caller). If not, a "no call" is reported.
If an SNV call is within `mindist` base pairs of the end of the
alignment it is ignored. Set to `None` to disable this behavior.
Variant calls with no spanning interesting k-mers are designated as
"passenger calls" and discarded.
"""
offset = 0 if self.targetshort else self.offset
if self.vartype == 'snv':
caller = self.call_snv(self.match.query,
self.match.target,
offset,
ksize,
mindist,
logstream=logstream)
for call in caller:
if self.is_passenger(call):
call.filter(vf.PassengerVariant)
yield call
elif self.vartype == 'indel':
indelcaller = self.call_indel(ksize)
indel = next(indelcaller)
if self.is_passenger(indel):
indel.filter(vf.PassengerVariant)
yield indel
leftflankcaller = self.call_snv(self.leftflank.query,
self.leftflank.target,
offset,
ksize,
mindist,
donocall=False)
offset += self.leftflank.length
if self.indeltype == 'D':
offset += self.indel.length
rightflankcaller = self.call_snv(self.rightflank.query,
self.rightflank.target,
offset,
ksize,
mindist,
donocall=False)
for call in chain(leftflankcaller, rightflankcaller):
if self.is_passenger(call):
call.filter(vf.PassengerVariant)
yield call
else:
nocall = Variant(self.seqid,
self.pos,
'.',
'.',
CONTIG=self.varseq,
CIGAR=self.cigar,
KSW2=str(self.score))
nocall.filter(vf.InscrutableCigar)
yield nocall
def call_snv(self,
qseq,
tseq,
offset,
ksize,
mindist=6,
donocall=True,
logstream=sys.stderr):
"""Call SNVs from the aligned mismatched sequences.
The `qseq` and `tseq` are strings containing query and target sequences
of identical length; `mismatches` is a list of positions where `qseq`
and `tseq` do not match; `offset` is the number of 5' nucleotides in
the target not aligned to the query; and `ksize` is used to compute a
window that spans all reference allele k-mers in `tseq` and all
alternate allele k-mers in `qseq`.
"""
length = len(qseq)
assert len(tseq) == length
diffs = [i for i in range(length) if tseq[i] != qseq[i]]
if mindist:
diffs = trim_terminal_snvs(diffs, length, mindist, logstream)
if len(diffs) == 0:
if donocall:
nocall = Variant(self.seqid,
self.cutout.local_to_global(offset),
'.',
'.',
CONTIG=qseq,
CIGAR=self.cigar,
KSW2=str(self.score),
IKMERS=str(len(self.contig.annotations)))
nocall.filter(vf.PerfectMatch)
yield nocall
return
for pos in diffs:
minpos = max(pos - ksize + 1, 0)
maxpos = min(pos + ksize, length)
altwindow = qseq[minpos:maxpos]
refrwindow = tseq[minpos:maxpos]
refr = tseq[pos].upper()
alt = qseq[pos].upper()
localcoord = pos + offset
globalcoord = self.cutout.local_to_global(localcoord)
nikmers = n_ikmers_present(self.contig, altwindow)
snv = Variant(self.seqid,
globalcoord,
refr,
alt,
CONTIG=qseq,
CIGAR=self.cigar,
KSW2=str(self.score),
IKMERS=str(nikmers),
ALTWINDOW=altwindow,
REFRWINDOW=refrwindow)
yield snv
def test_filter_field():
v = Variant('scaffold1', 12345, '.', '.')
assert v.filterstr == '.'
v.filter(vf.InscrutableCigar)
assert v.filterstr == 'InscrutableCigar'
v = Variant('chr1', 55555, '.', '.')
v.filter(vf.PerfectMatch)
assert v.filterstr == 'PerfectMatch'
v = Variant('1', 809768, 'C', 'CAT')
assert v.filterstr == 'PASS'
v.filter(vf.PassengerVariant)
assert v.filterstr == 'PassengerVariant'
v.filter(vf.Homopolymer)
assert v.filterstr == 'Homopolymer;PassengerVariant'
v = Variant('one', 112358, 'T', 'A')
v.filter('SNPyMcSNPface')
v.filter(6.022e23)
v.filter(dict(chicken='waffles', biscuits='gravy'))
v.filterstr == 'PASS' # These "filters" shouldn't actually do anything