def parseInsertion(record, dataHub):
altchars = set(record.alt.upper())
breakpoint = utilities.Locus(record.chrom, record.start, record.end, "+")
if altchars <= set("ACGTN"):
insertSeq = record.alt.upper()
variant = variants.Insertion(breakpoint, insertSeq, dataHub.alignDistance, dataHub.genome)
elif "MEINFO" in record.info:
meinfo = record.info["MEINFO"].split(",")
meName = meinfo[0]
meStart = utilities.getListDefault(meinfo, 1, 0)
meEnd = utilities.getListDefault(meinfo, 2, 1e100)
meStrand = utilities.getListDefault(meinfo, 3, "+")
meLocus = utilities.Locus(meName, meStart, meEnd, meStrand)
variant = variants.MobileElementInsertion(breakpoint, meLocus, getMobileElementFasta(dataHub),
dataHub.alignDistance, dataHub.genome)
else:
raise VCFParserError("Unknown insertion sequence")
return variant
def parseInsertion(record, dataHub):
altchars = set(record.alt.upper())
breakpoint = utilities.Locus(record.chrom, record.start, record.end, "+")
if altchars <= set("ACGTN"):
insertSeq = record.alt.upper()
variant = variants.Insertion(breakpoint, insertSeq,
dataHub.alignDistance, dataHub.genome)
elif "MEINFO" in record.info:
meinfo = record.info["MEINFO"].split(",")
meName = meinfo[0]
meStart = utilities.getListDefault(meinfo, 1, 0)
meEnd = utilities.getListDefault(meinfo, 2, 1e100)
meStrand = utilities.getListDefault(meinfo, 3, "+")
meLocus = utilities.Locus(meName, meStart, meEnd, meStrand)
variant = variants.MobileElementInsertion(
breakpoint, meLocus, getMobileElementFasta(dataHub),
dataHub.alignDistance, dataHub.genome)
else:
raise VCFParserError("Unknown insertion sequence")
return variant
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
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