def setArgs(self, args):
self.args = args
try:
self.genome = genomesource.FastaGenomeSource(args.ref)
for bamPath in self.args.bam:
name = nameFromBamPath(bamPath)
i = 0
while name in self.samples:
i += 1
curname = "{}_{}".format(name, i)
if curname not in self.samples:
name = curname
break
sample = Sample(name, bamPath)
self.samples[name] = sample
if self.args.annotations:
for annoPath in self.args.annotations:
name = nameFromBedPath(annoPath)
if annoPath.endswith(".bed") or annoPath.endswith(
".bed.gz"):
self.annotationSets[name] = annotations.AnnotationSet(
annoPath)
else:
if not (annoPath.endswith(".gff") or annoPath.endswith(".gff.gz") \
or annoPath.endswith(".gtf") or annoPath.endswith(".gtf.gz")):
logging.warn(
"Unknown annotation file extension; trying to parse as if GTF/GFF format: '{}'"
.format(annoPath))
self.annotationSets[name] = gff.GeneAnnotationSet(
annoPath)
except:
self.args._parser.print_help()
print("")
raise
for bamPath in self.args.bam:
try:
bam = pysam.AlignmentFile(bamPath)
bam.fetch()
except ValueError:
logging.error(
"\nERROR: Need to create index for input bam file: {}".
format(bamPath))
sys.exit(0)
def getVariant(dataHub):
if dataHub.args.type.lower().startswith("del"):
assert len(dataHub.args.breakpoints) == 3, getBreakpointFormatsStr("del")
chrom = dataHub.args.breakpoints[0]
start = int(dataHub.args.breakpoints[1])
end = int(dataHub.args.breakpoints[2])
assert start < end
variant = Deletion.from_breakpoints(chrom, start-1, end-1, dataHub.alignDistance, dataHub.genome)
elif dataHub.args.type.lower() in ["ldel", "largedeletion"]:
assert len(dataHub.args.breakpoints) == 3, getBreakpointFormatsStr("ldel")
chrom = dataHub.args.breakpoints[0]
start = int(dataHub.args.breakpoints[1])
end = int(dataHub.args.breakpoints[2])
assert start < end
variant = LargeDeletion.from_breakpoints(chrom, start-1, end-1, dataHub.alignDistance, dataHub.genome)
elif dataHub.args.type.lower().startswith("ins"):
assert len(dataHub.args.breakpoints) in [3,4], getBreakpointFormatsStr("ins")
chrom = dataHub.args.breakpoints[0]
pos = int(dataHub.args.breakpoints[1])
if len(dataHub.args.breakpoints) == 3:
seq = dataHub.args.breakpoints[2]
end = pos
else:
end = int(dataHub.args.breakpoints[2])
seq = dataHub.args.breakpoints[3]
variant = Insertion(Locus(chrom, pos, end, "+"), seq, dataHub.alignDistance, dataHub.genome)
elif dataHub.args.type.lower().startswith("inv"):
assert len(dataHub.args.breakpoints) == 3, getBreakpointFormatsStr("inv")
chrom = dataHub.args.breakpoints[0]
start = int(dataHub.args.breakpoints[1])
end = int(dataHub.args.breakpoints[2])
if dataHub.args.min_mapq is None:
dataHub.args.min_mapq = -1
variant = Inversion(Locus(chrom, start, end, "+"), dataHub.alignDistance, dataHub.genome)
elif dataHub.args.type.lower().startswith("mei"):
assert len(dataHub.args.breakpoints) >= 4, getBreakpointFormatsStr("mei")
insertionBreakpoint = Locus(dataHub.args.breakpoints[1], dataHub.args.breakpoints[2], dataHub.args.breakpoints[2], "+")
meName = dataHub.args.breakpoints[3]
meStrand = getListDefault(dataHub.args.breakpoints, 4, "+")
meStart = getListDefault(dataHub.args.breakpoints, 5, 0)
meEnd = getListDefault(dataHub.args.breakpoints, 6, 1e100)
meCoords = Locus(meName, meStart, meEnd, meStrand)
meFasta = genomesource.FastaGenomeSource(dataHub.args.breakpoints[0])
variant = MobileElementInsertion(insertionBreakpoint, meCoords, meFasta, dataHub.alignDistance, dataHub.genome)
elif dataHub.args.type.lower().startswith("tra"):
assert len(dataHub.args.breakpoints) == 5, getBreakpointFormatsStr("tra")
chrom1 = dataHub.args.breakpoints[0]
start1 = int(dataHub.args.breakpoints[1])
chrom2 = dataHub.args.breakpoints[2]
start2 = int(dataHub.args.breakpoints[3])
orientation = dataHub.args.breakpoints[4]
if dataHub.args.min_mapq is None:
dataHub.args.min_mapq = -1
variant = Translocation(Locus(chrom1, start1, start1, "+"),
Locus(chrom2, start2, start2, orientation),
dataHub.alignDistance, dataHub.genome)
elif dataHub.args.type.lower() in ["bkend", "breakend"]:
assert len(dataHub.args.breakpoints) == 6, getBreakpointFormatsStr("bkend")
chrom1 = dataHub.args.breakpoints[0]
start1 = int(dataHub.args.breakpoints[1])
strand1 = dataHub.args.breakpoints[2]
chrom2 = dataHub.args.breakpoints[3]
start2 = int(dataHub.args.breakpoints[4])
strand2 = dataHub.args.breakpoints[5]
if dataHub.args.min_mapq is None:
dataHub.args.min_mapq = -1
variant = Breakend(Locus(chrom1, start1, start1, strand1),
Locus(chrom2, start2, start2, strand2),
dataHub.alignDistance, dataHub.genome)
else:
raise Exception("only accept event types of deletion, insertion, mei, translocation or breakend")
logging.info(" Variant: {}".format(variant))
return variant