def polishSegment(self, seg, als):
# type: (Segment, List[AlignmentPiece]) -> AlignmentPiece
sys.stdout.trace("Polishing segment", seg)
w = max(900, params.k)
r = 50
first = seg.left / w
last = min(seg.right + w - 1, len(seg.contig)) / w
segs = []
for i in range(first, last + 1):
segs.append(
Segment(seg.contig, max(0, i * w - r),
min(len(seg.contig), (i + 1) * w + r)))
als_by_segment = [[] for i in range(last - first + 1)]
for al in als:
l = al.seg_to.left / w
r = al.seg_to.right / w + 1
for i in range(max(0, l - first - 1),
min(last - first + 1, r - first + 2)):
if al.seg_to.inter(segs[i]):
als_by_segment[i].append(al)
res_als = []
for seg1, seg_als in zip(segs, als_by_segment):
if seg1.inter(seg):
res_als.append(self.polishSmallSegment(seg1, seg_als))
res = AlignmentPiece.GlueOverlappingAlignments(res_als)
return res.reduce(target=seg)
def testManual(self):
contig1 = Contig("ACGTACGTA", "from")
contig2 = Contig("ACTACGTACGTACAT", "to")
al1 = AlignmentPiece(contig1.asSegment(), contig2.segment(0, 8),
"2M1I6M")
al2 = AlignmentPiece(contig1.segment(0, 8), contig2.segment(7, 15),
"8M")
glued = AlignmentPiece.GlueOverlappingAlignments([al1, al2])
assert glued.cigar == "2M1I5M8M", str(glued) + " " + glued.cigar
assert glued.seg_from.Seq(
) == "ACGTACGTACGTACGT", str(glued) + " " + glued.cigar
assert al1.reduce(query=contig1.segment(0, 2)).cigar == "2M"
assert al1.reduce(query=contig1.segment(0, 3)).cigar == "2M"
assert al1.reduce(query=contig1.segment(0, 4)).cigar == "2M1I1M"
