def chainChromTarget(target, hal, queryGenome, queryChromSize, queryTwoBit, queryChrom, targetGenome, targetTwoBit, chainFile):
"target to chain one chromosome"
queryBed = makeChromBedTmp(target, hal, queryChrom, queryChromSize)
# --inMemory caused out of memory in with some alignments with 31G allocated
procOps.callProc([["halLiftover", "--outPSL", hal, queryGenome, queryBed, targetGenome, "/dev/stdout"],
["pslPosTarget", "/dev/stdin", "/dev/stdout"],
["axtChain", "-psl", "-verbose=0", "-linearGap=medium", "/dev/stdin", targetTwoBit, queryTwoBit, chainFile]])
def chainChromTarget(target, hal, queryGenome, queryChromSize, queryTwoBit, queryChrom, targetGenome, targetTwoBit, chainFile):
"target to chain one chromosome"
queryBed = makeChromBedTmp(target, hal, queryChrom, queryChromSize)
# --inMemory caused out of memory in with some alignments with 31G allocated
procOps.callProc([["halLiftover", "--outPSL", hal, queryGenome, queryBed, targetGenome, "/dev/stdout"],
["pslPosTarget", "/dev/stdin", "/dev/stdout"],
["axtChain", "-psl", "-verbose=0", "-linearGap=medium", "/dev/stdin", targetTwoBit, queryTwoBit, chainFile]])
def pileup(out_bam, vcf_path):
bases = {"A", "T", "G", "C", "a", "t", "g", "c"}
vcf_handle = vcf.Reader(open(vcf_path))
wgs_results = defaultdict(list)
for vcf_rec in vcf_handle:
if vcf_rec.is_indel:
continue
pos_str = "{0}:{1}-{1}".format(vcf_rec.CHROM, vcf_rec.POS)
cmd = ['samtools', 'mpileup', '-r', pos_str, out_bam]
mpileup_rec = callProc(cmd).split()
pile_up_result = Counter(x.upper() for x in mpileup_rec[4]
if x in bases)
sample_dict = {s.sample: s.gt_bases for s in vcf_rec.samples}
for s in vcf_rec.samples:
if len([x for x in sample_dict.itervalues() if x == s.gt_bases
]) != 1:
continue
if s.gt_bases is None:
continue
c = 1.0 * pile_up_result[s.gt_bases] / len(mpileup_rec[4])
c *= 1.0 * len([
x for x in sample_dict.itervalues() if x is not None
]) / len(sample_dict)
wgs_results[s.sample].append([vcf_rec.POS, c])
return wgs_results
def pileup(out_bam, vcf_path):
bases = {"A", "T", "G", "C", "a", "t", "g", "c"}
vcf_handle = vcf.Reader(open(vcf_path))
wgs_results = defaultdict(list)
for vcf_rec in vcf_handle:
if vcf_rec.is_indel:
continue
pos_str = "{0}:{1}-{1}".format(vcf_rec.CHROM, vcf_rec.POS)
cmd = ['samtools', 'mpileup', '-r', pos_str, out_bam]
mpileup_rec = callProc(cmd).split()
pile_up_result = Counter(x.upper() for x in mpileup_rec[4] if x in bases)
sample_dict = {s.sample: s.gt_bases for s in vcf_rec.samples}
for s in vcf_rec.samples:
if len([x for x in sample_dict.itervalues() if x == s.gt_bases]) != 1:
continue
if s.gt_bases is None:
continue
c = 1.0 * pile_up_result[s.gt_bases] / len(mpileup_rec[4])
c *= 1.0 * len([x for x in sample_dict.itervalues() if x is not None]) / len(sample_dict)
wgs_results[s.sample].append([vcf_rec.POS, c])
return wgs_results
def testCallSimple(self):
out = procOps.callProc(["sort", self.getInputFile("simple1.txt")])
self.assertEqual(out, "five\nfour\none\nsix\nthree\ntwo")
def testCallErr(self):
with self.assertRaises(pipeline.ProcException) as cm:
procOps.callProc(["false"])
self.assertEqual(str(cm.exception), 'process exited 1: false')
def testCallKeepNL(self):
out = procOps.callProc(["sort", self.getInputFile("simple1.txt")], keepLastNewLine=True)
self.assertEqual(out, "five\nfour\none\nsix\nthree\ntwo\n")
def testCallSimple(self):
out = procOps.callProc(["sort", self.getInputFile("simple1.txt")])
self.failUnlessEqual(out, "five\nfour\none\nsix\nthree\ntwo")
def testQuotesCsh(self):
# must use /bin/echo, as some csh echos expand backslash sequences
for (words, expect) in self.testData:
out = procOps.callProc(
["csh", "-c", "/bin/echo " + " ".join(procOps.shQuote(words))])
self.failUnlessEqual(out, expect)
def testQuotesBash(self):
for (words, expect) in self.testData:
out = procOps.callProc(
["sh", "-c", "/bin/echo " + " ".join(procOps.shQuote(words))])
self.failUnlessEqual(out, expect)
def makeChromSizeTmp(target, prefix, twoBit):
"create a tmp chrom sizes file from a twobit"
chromSizeFile = getLocalTempPath(target, prefix, "size")
procOps.callProc(["twoBitInfo", twoBit, chromSizeFile])
return chromSizeFile
def testCallKeepNL(self):
out = procOps.callProc(
["sort", self.getInputFile("simple1.txt")], keepLastNewLine=True)
self.failUnlessEqual(out, "five\nfour\none\nsix\nthree\ntwo\n")
def makeChromSizeTmp(target, prefix, twoBit):
"create a tmp chrom sizes file from a twobit"
chromSizeFile = getLocalTempPath(target, prefix, "size")
procOps.callProc(["twoBitInfo", twoBit, chromSizeFile])
return chromSizeFile
def testCallErr(self):
ex = None
try:
procOps.callProc(["false"])
except Exception, ex:
pass
def testQuotesBash(self):
for (words, expect) in self.testData:
out = procOps.callProc(["sh", "-c", "/bin/echo " + " ".join(procOps.shQuote(words))])
self.assertEqual(out, expect)
def testQuotesCsh(self):
# must use /bin/echo, as some csh echos expand backslash sequences
for (words, expect) in self.testData:
out = procOps.callProc(["csh", "-c", "/bin/echo " + " ".join(procOps.shQuote(words))])
self.assertEqual(out, expect)
def testCallErr(self):
ex = None
try:
procOps.callProc(["false"])
except Exception, ex:
pass
