def getSeq(self, chrom, start, end, strand):
chrom = matchChromFormat(chrom, self.fasta.keys())
seq = self.fasta[chrom][start:end+1]
if strand == "-":
seq = reverseComp(seq)
return seq
def alignBothStrands(seq, aligner): #, target):
revseq = reverseComp(seq)
forward_al = aligner.align(seq)
reverse_al = aligner.align(revseq)
strand = None
if not forward_al:
if reverse_al:
strand = "-"
else:
if not reverse_al:
strand = "+"
else:
if forward_al.score >= reverse_al.score:
strand = "+"
if reverse_al.score > forward_al.score2:
forward_al.score2 = reverse_al.score
forward_al.ref_end2 = -1
else:
strand = "-"
if forward_al.score > reverse_al.score2:
reverse_al.score2 = forward_al.score
reverse_al.ref_end2 = -1
if strand == "+":
return "+", forward_al
else:
return "-", reverse_al
def alignBothStrands(seq, aligner):
revseq = reverseComp(seq)
forward_al = aligner.align(seq)
reverse_al = aligner.align(revseq)
strand = None
if not forward_al:
if reverse_al:
strand = "-"
else:
if not reverse_al:
strand = "+"
else:
if forward_al.score >= reverse_al.score:
strand = "+"
if reverse_al.score > forward_al.score2:
forward_al.score2 = reverse_al.score
forward_al.ref_end2 = -1
else:
strand = "-"
if forward_al.score > reverse_al.score2:
reverse_al.score2 = forward_al.score
reverse_al.ref_end2 = -1
if strand == "+":
return "+", forward_al
else:
return "-", reverse_al
def getSeq(self, chrom, start, end, strand):
chrom = matchChromFormat(chrom, list(self.fasta.keys()))
seq = self.fasta[chrom][start:end+1]
if strand == "-":
seq = reverseComp(seq)
return seq
def remap(self, seq):
results = {}
for name, aligner in self.namesToAligners.items():
results[name] = None
if self.tryExact:
revseq = reverseComp(seq)
results[name] = tryAlignExact(seq, revseq, self.namesToRefs[name], aligner)
if results[name] is None:
results[name] = alignBothStrands(seq, aligner)#, self.namesToRefs[name])
return seq, results
def remap(self, seq):
results = {}
for name, aligner in self.namesToAligners.items():
results[name] = None
if self.tryExact:
revseq = reverseComp(seq)
results[name] = tryAlignExact(seq, revseq,
self.namesToRefs[name], aligner)
if results[name] is None:
results[name] = alignBothStrands(
seq, aligner) #, self.namesToRefs[name])
return seq, results
def chooseBestAlignment(read, mappings, chromPartsCollection):
# TODO: this is kind of ridiculous; we need to make pickleable reads that can be sent to
# and from the Multimapper
# mappings: name -> (strand, aln)
bestName = None
bestAln = None
bestStrand = None
secondScore = None
for name, mapping in mappings.iteritems():
strand, aln = mapping
if bestAln is None or aln.score > bestAln.score:
bestName = name
bestAln = aln
bestStrand = strand
for name, mapping in mappings.iteritems():
strand, aln = mapping
if name == bestName:
if secondScore is None or bestAln.score2 > secondScore:
secondScore = bestAln.score2
else:
if secondScore is None or aln.score > secondScore:
secondScore = aln.score
seq = read.seq
genome_seq = chromPartsCollection.getPart(bestName).getSeq()[bestAln.ref_begin : bestAln.ref_end + 1].upper()
if bestStrand == "-":
seq = reverseComp(seq)
if read.is_reverse:
bestStrand = "+" if bestStrand == "-" else "-"
bestAln = Alignment(
read.qname,
bestName,
bestAln.ref_begin,
bestAln.ref_end,
bestStrand,
seq,
bestAln.cigar_string,
bestAln.score,
genome_seq,
secondScore,
read.mapq,
)
return bestAln
def chooseBestAlignment(read, mappings, chromPartsCollection):
# TODO: this should be read-pair aware
# TODO: this is kind of ridiculous; we need to make pickleable reads that can be sent to
# and from the Multimapper
# mappings: name -> (strand, aln)
bestName = None
bestAln = None
bestStrand = None
secondScore = None
for name, mapping in mappings.items():
if mapping is None:
return None
strand, aln = mapping
if bestAln is None or aln.score > bestAln.score:
bestName = name
bestAln = aln
bestStrand = strand
for name, mapping in mappings.items():
strand, aln = mapping
if name == bestName:
if secondScore is None or bestAln.score2 > secondScore:
secondScore = bestAln.score2
else:
if secondScore is None or aln.score > secondScore:
secondScore = aln.score
seq = read.seq
genome_seq = chromPartsCollection.getPart(
bestName).getSeq()[bestAln.ref_begin:bestAln.ref_end + 1].upper()
if bestStrand == "-":
seq = reverseComp(seq)
if read.is_reverse:
bestStrand = "+" if bestStrand == "-" else "-"
bestAln = Alignment(read.qname, bestName, bestAln.ref_begin,
bestAln.ref_end, bestStrand, seq, bestAln.cigar_string,
bestAln.score, genome_seq, secondScore, read.mapq)
return bestAln
def findBestAlignment(seq, aligner):
revseq = reverseComp(seq)
forward_al = aligner.align(seq)
reverse_al = aligner.align(revseq)
strand = None
if not forward_al:
if reverse_al:
strand = "-"
else:
if not reverse_al:
strand = "+"
else:
if forward_al.score >= reverse_al.score:
strand = "+"
else:
strand = "-"
if strand == "+":
return "+", forward_al
else:
return "-", reverse_al
def getSeq(self, chrom, start, end, strand):
seq = self.seq[start:end+1]
if strand == "-":
seq = reverseComp(seq)
return seq
def getSeq(self, chrom, start, end, strand):
seq = self.fasta[chrom][start:end + 1]
if strand == "-":
seq = reverseComp(seq)
return seq