def makeVCFline(scaffold,
position,
GTdict,
names,
refDict=None,
genoFormat=None):
genomeSite = genomics.GenomeSite(genoDict=GTdict,
sampleNames=names,
genoFormat=genoFormat)
alleles = genomeSite.alleles(byFreq=True)
if alleles == []: alleles = ["N"]
if refDict:
refBase = refDict[scaffold][int(position) - 1]
if refBase in alleles: alleles.pop(alleles.index(refBase))
alleles = [refBase] + alleles
else:
refBase = alleles[0]
alt = alleles[1:]
if alt == []: alt = ["."]
codedGenos = genomeSite.asList(mode="coded", alleles=alleles)
output = [
scaffold,
str(position), ".", refBase, ",".join(alt), ".", ".", ".", "GT"
] + codedGenos
return "\t".join(output)
def getPopIndBaseCounts(siteData, genoFormat, allSamples, popDict, ploidyDict):
site = genomics.GenomeSite(genotypes=[siteData["GTs"][name] for name in allSamples], sampleNames=allSamples,
popDict=popDict, ploidyDict=ploidyDict, genoFormat=args.genoFormat)
popIndBaseCounts = dict([(popName, np.array([site.genotypes[indName].asBaseCounts() for indName in popDict[popName]]),) for popName in popNames])
return popIndBaseCounts
def analysisWrapper(inQueue, outQueue, inputGenoFormat, outputGenoFormat,
headers, include, exclude, samples, minCalls, minPopCalls,
minAlleles, maxAlleles, minVarCount, maxHet, minFreq,
maxFreq, HWE_P, HWE_side, popDict, ploidyDict, fixed,
nearlyFixedDiff, forcePloidy, thinDist, noTest):
sampleIndices = [headers.index(s) for s in samples]
while True:
podNumber, inPod = inQueue.get()
if verbose:
print >> sys.stderr, "Pod", podNumber, "received for analysis."
outPod = []
lastScaf = None
for lineData in inPod:
lineNumber, line = lineData
#if verbose: print >> sys.stderr, "Analysing line", lineNumber
objects = line.split()
if (include and objects[0]
not in include) or (exclude and objects[0] in exclude):
continue
site = genomics.GenomeSite(
genotypes=[objects[i] for i in sampleIndices],
sampleNames=samples,
popDict=popDict,
ploidyDict=ploidyDict,
genoFormat=inputGenoFormat,
forcePloidy=forcePloidy)
goodSite = True
if thinDist:
pos = int(objects[1])
if lastScaf != objects[0]:
lastPos = pos
lastScaf = objects[0]
goodSite = False
elif pos - lastPos < thinDist:
goodSite = False
if goodSite and not noTest:
goodSite = genomics.siteTest(site,
samples=samples,
minCalls=minCalls,
minPopCalls=minPopCalls,
minAlleles=minAlleles,
maxAlleles=maxAlleles,
minVarCount=minVarCount,
maxHet=maxHet,
minFreq=minFreq,
maxFreq=maxFreq,
HWE_P=HWE_P,
HWE_side=HWE_side,
fixed=fixed,
nearlyFixedDiff=nearlyFixedDiff)
if goodSite:
outLine = "\t".join(objects[:2] + [
str(g) for g in site.asList(samples, mode=outputGenoFormat)
]) + "\n"
outPod.append((lineNumber, outLine))
if thinDist: lastPos = int(objects[1])
#if verbose: print >> sys.stderr, objects[0], objects[1], "passed: ", goodSite
outQueue.put((podNumber, outPod))
if verbose:
print >> sys.stderr, "Pod", podNumber, "analysed, sent to sorter."
continue
#if there are intervals, check whether the site matches any
if intervalsFile:
siteIntervals = whichInterval(siteData["scaffold"],
siteData["position"], scafIntervals,
intervalPosDict)
else:
siteIntervals = [0]
if not siteIntervals: continue
site = genomics.GenomeSite(
genotypes=[siteData["GTs"][name] for name in allSamples],
sampleNames=allSamples,
popDict=popDict,
ploidyDict=ploidyDict,
genoFormat=args.genoFormat)
popIndBaseCounts = dict([(
popName,
np.array([
site.genotypes[indName].asBaseCounts()
for indName in popDict[popName]
]),
) for popName in popNames])
# get population basec counts, and do the subsampling if necessary
# This is currently conservative. If any one of the populations lacks sufficient good genotypes it will break
# in theory we could modify this part to use info for the pops it can - might be necessary when sites are limited
and objects[0] in exclude):
continue
#if there are intervals, check whether the site matches any
if intervalsFile:
siteIntervals = whichInterval(objects[0], int(objects[1]),
scafIntervals, intervalPosDict)
else:
siteIntervals = [0]
if not siteIntervals: continue
site = genomics.GenomeSite(genotypes=[objects[i] for i in sampleIndices],
sampleNames=allSamples,
popDict=popDict,
ploidyDict=ploidyDict,
genoFormat=args.genoFormat)
popIndBaseCounts = dict([(
popName,
np.array([
site.genotypes[indName].asBaseCounts()
for indName in popDict[popName]
]),
) for popName in popNames])
# get population basec counts, and do the subsampling if necessary
# This is currently conservative. If any one of the populations lacks sufficient good genotypes it will break
# in theory we could modify this part to use info for the pops it can - might be necessary when sites are limited
linesDone = 0
scaf = None
chrom = None
pos = 0
if args.cumulativePos:
#dict giving the last known position of each chrom from the previous scaffold
chromOffset = dict(zip(chromDict.keys(), [0] * len(chromDict)))
chromOffset[str(args.nullChrom)] = 0
for line in genoFile:
site = genomics.parseGenoLine(line)
genomeSite = genomics.GenomeSite(genotypes=site.GTs,
sampleNames=allNames,
genoFormat=args.genoFormat)
if len(genomeSite.alleles()) == 2:
counts = genomeSite.asList(mode="count", samples=samples, missing=9)
genoOut.write("".join([str(c) for c in counts]) + "\n")
if site.scaffold != scaf:
#different scaffold from the last site
#if using cumulative positions, change the offset for the last chrom
if chrom is not None and args.cumulativePos:
chromOffset[chrom] = pos
#now get new scaf, chrom and pos
scaf = site.scaffold
try:
linesDone = 0
scaf = None
chrom = None
pos = 0
if args.cumulativePos:
#dict giving the last known position of each chrom from the previous scaffold
chromOffset = dict(zip(chromDict.keys(), [0]*len(chromDict)))
chromOffset[str(args.nullChrom)] = 0
for siteData in reader.siteBySite():
genomeSite = genomics.GenomeSite(genoDict = siteData["GTs"], genoFormat=args.genoFormat)
alleles = genomeSite.alleles()
if len(alleles) == 2:
counts = genomeSite.asList(mode="count", samples=samples, missing = 9)
genoOut.write("".join([str(c) for c in counts]) + "\n")
if siteData["scaffold"] != scaf:
#different scaffold from the last site
#if using cumulative positions, change the offset for the last chrom
if chrom is not None and args.cumulativePos: chromOffset[chrom] = pos
#now get new scaf, chrom and pos
scaf = siteData["scaffold"]
try: chrom = chromDict[scaf]
def analysisWrapper(inQueue, outQueue, inputGenoFormat, headers, include,
exclude, group1inds, group2inds, permutations,
permutationMaxP):
samples = group1inds + group2inds
sampleIndices = [headers.index(s) for s in samples]
group1 = np.array([True] * len(group1inds) + [False] * len(group2inds))
group2 = ~group1
while True:
podNumber, inPod = inQueue.get()
if verbose:
sys.stderr.write(
"Pod {} received for analysis...\n".format(podNumber))
outPod = []
for lineData in inPod:
lineNumber, line = lineData
#if verbose: print >> sys.stderr, "Analysing line", lineNumber
objects = line.split()
if (include and objects[0]
not in include) or (exclude and objects[0] in exclude):
continue
site = genomics.GenomeSite(
genotypes=[objects[i] for i in sampleIndices],
sampleNames=samples,
genoFormat=inputGenoFormat)
alleles = site.alleles()
if len(alleles) == 2:
minorCount = np.array(
site.asList(mode="count",
countAllele=alleles[1],
missing=-1))
majorCount = np.array(
site.asList(mode="count",
countAllele=alleles[0],
missing=-1))
#get index for good genotypes and filter all by that
idx = np.where(minorCount >= 0)[0]
_group1_ = group1[idx]
_group2_ = group2[idx]
minorPresent = minorCount[idx] >= 1
minorAbsent = ~minorPresent
majorPresent = majorCount[idx] >= 1
majorAbsent = ~majorPresent
minorTable = np.array([[(minorPresent & _group1_).sum(),
(minorAbsent & _group1_).sum()],
[(minorPresent & _group2_).sum(),
(minorAbsent & _group2_).sum()]])
majorTable = np.array([[(majorPresent & _group1_).sum(),
(majorAbsent & _group1_).sum()],
[(majorPresent & _group2_).sum(),
(majorAbsent & _group2_).sum()]])
p_values = (
fisher_exact(minorTable)[1],
fisher_exact(majorTable)[1],
)
result = [min(p_values)]
if permutations >= 1:
if permutationMaxP is None or result[0] <= permutationMaxP:
table = minorTable if p_values[0] <= p_values[
1] else majorTable
phi = chisquare(table, axis=None)[0] / table.sum()
phi_permuted = []
for i in range(permutations):
newGroup1 = np.random.permutation(_group1_)
newGroup2 = ~newGroup1
newTable = np.array(
[[(minorPresent & newGroup1).sum(),
(minorAbsent & newGroup1).sum()],
[(minorPresent & newGroup2).sum(),
(minorAbsent & newGroup2).sum()]])
phi_permuted.append(
chisquare(newTable, axis=None)[0] /
table.sum())
p_emp = (len(
[_phi_
for _phi_ in phi_permuted if _phi_ >= phi]) +
1.) / (permutations + 1.)
else:
p_emp = np.NaN
result.append(p_emp)
elif permutations >= 1:
result = [np.NaN] * 2
else:
result = [np.NaN]
outLine = "\t".join(objects[:2] +
[str(round(x, 5)) for x in result]) + "\n"
outPod.append((lineNumber, outLine))
outQueue.put((podNumber, outPod))
if verbose:
sys.stderr.write(
"Pod {} analysed, sent to sorter.\n".format(podNumber))
