def map_isoforms_to_reference_transcripts(self):
"""Map isoforms to reference transcripts."""
m5out = self.output_analysis_fn + ".blasr.out.m5"
cmd = 'blasr %s %s --bestn 1 -m 5 --out %s' % \
(self.isoseq_output_fa, self.reference_transcripts_fn, m5out)
execute(cmd)
return [r for r in BLASRM5Reader(m5out)]
def blasr_against_ref2(output_filename,
is_FL,
sID_starts_with_c,
qver_get_func,
qvmean_get_func,
qv_prob_threshold=.03,
ece_penalty=1,
ece_min_len=20,
same_strand_only=True,
max_missed_start=200,
max_missed_end=50,
full_missed_start=50,
full_missed_end=30):
"""
Excluding criteria:
(1) self hit
(2) opposite strand hit (should already be in the same orientation;
can override with <same_strand_only> set to False)
(3) less than 90% aligned or more than 50 bp missed
qver_get_func --- should be basQV.basQVcacher.get() or
.get_smoothed(), or can just pass in
lambda (x, y): 1. to ignore QV
"""
with BLASRM5Reader(output_filename) as reader:
for r in reader:
missed_q = r.qStart + r.qLength - r.qEnd
missed_t = r.sStart + r.sLength - r.sEnd
if sID_starts_with_c:
# because all consensus should start with
# c<cluster_index>
assert r.sID.startswith('c')
if r.sID.find('/') > 0:
r.sID = r.sID.split('/')[0]
if r.sID.endswith('_ref'):
# probably c<cid>_ref
cID = int(r.sID[1:-4])
else:
cID = int(r.sID[1:])
else:
cID = r.sID
# self hit, useless!
# opposite strand not allowed!
if (cID == r.qID or (r.strand == '-' and same_strand_only)):
yield HitItem(qID=r.qID, cID=cID)
continue
# regardless if whether is full-length (is_FL)
# the query MUST be mapped fully (based on full_missed_start/end)
if r.qStart > full_missed_start or (r.qLength -
r.qEnd) > full_missed_end:
yield HitItem(qID=r.qID, cID=cID)
# full-length case: allow up to max_missed_start bp of 5' not aligned
# and max_missed_end bp of 3' not aligned
# non-full-length case: not really tested...don't use
if is_FL and not alignment_missed_start_end_less_than_threshold(r,\
max_missed_start, max_missed_end, full_missed_start, full_missed_end):
yield HitItem(qID=r.qID, cID=cID)
else:
cigar_str, ece_arr = eval_blasr_alignment(
record=r,
qver_get_func=qver_get_func,
qvmean_get_func=qvmean_get_func,
sID_starts_with_c=sID_starts_with_c,
qv_prob_threshold=qv_prob_threshold)
if alignment_has_large_nonmatch(ece_arr, ece_penalty,
ece_min_len):
yield HitItem(qID=r.qID, cID=cID)
else:
yield HitItem(qID=r.qID,
cID=cID,
qStart=r.qStart,
qEnd=r.qEnd,
missed_q=missed_q * 1. / r.qLength,
missed_t=missed_t * 1. / r.sLength,
fakecigar=cigar_str,
ece_arr=ece_arr)