def prepPop2(allInds, database, refPop, isXPEHH):
f = open(allInds)
w = open(allInds + "_", "w")
for l in f:
toks = l.rstrip().replace(" ", "\t").split("\t")
w.write(toks[1] + "\n")
w.close()
for chro in range(1, 23 + 1): # Human dependent
if not karyo.accept(chro, 1):
continue
if refPop != "shapeIt":
lexec.submit("python3",
"%s/beagle2ihs.py %s/%s/%s-%d.gz %s tmp-%d %s/37-%d.map %d"
% (MEGA.phasingScripts, MEGA.phaseDB, database,
refPop, chro, allInds, chro, MEGA.geneticMapDB, chro, chro))
else:
lexec.submit("python3",
"%s/beagle2ihs.py %s/%s/%d.gz %s tmp-%d %s/37-%d.map %d"
% (MEGA.phasingScripts, MEGA.phaseDB, database,
chro, allInds, chro, MEGA.geneticMapDB, chro, chro))
lexec.wait(True)
for chro in range(1, 23 + 1): # Human dependent
if not karyo.accept(chro, 1):
continue
if isXPEHH:
os.rename('tmp-%d.hap' % chro, '%d.hap' % chro)
os.rename('tmp-%d.map' % chro, '%d.map' % chro)
else:
removeFixed('tmp-%d' % chro, '%d' % chro)
removeFixed('tmp-%d' % chro, '%d' % chro)
if command == 'prepareData':
studyName = sys.argv[3]
allInds = sys.argv[4]
source = sys.argv[5]
refPop = sys.argv[6]
if len(sys.argv) > 7:
isXPEHH = True
else:
isXPEHH = False
prepPop2(allInds, source, refPop, isXPEHH)
elif command == 'iHS':
for k in range(1, maxChro + 1):
if not karyo.accept(k, 1): # needs whole chromosome accepted
continue
input = "%d.hap" % (k,)
inmap = "%d.map" % (k,)
out = "%d.uiHS" % (k,)
lexec.out = out
lexec.submit("ihs", "%s %s" % (inmap, input))
lexec.wait(True)
elif command == 'statIHS':
maxBin = sys.argv[3]
winSize = sys.argv[4]
minSNPs = sys.argv[5]
os.system("python3 %s/statIHS.py %s %s %s" % (
MEGA.haploScripts, maxBin, winSize, minSNPs))
end = pos
write_block(block, chro, start, end)
block += 1
return block
if args.step == "prepFST":
group = args.group
try:
os.mkdir(group)
except OSError:
pass
os.chdir(group)
os.system('rm gp-* 2>/dev/null')
plink.getIndivs("../" + indList, "fst", isBinary=False,
acceptFun=lambda chro, pos: chro == karyo.groups[group]
and karyo.accept(chro, pos), suf=myStudy.fs.sampSuff)
os.system("plink --noweb --recode --file fst --maf %f --out gp" %
myStudy.fs.MAF)
nblocks = slice_plink("gp", args.max)
for block in range(nblocks):
to_genepop("gp-%d" % block, "conv-%d" % block, popList)
elif args.step == "FST":
group = args.group
os.chdir(group)
for i in range(100000):
try:
os.mkdir(str(i))
except OSError:
pass
if os.path.exists("conv-%d.gp" % i):
os.chdir(str(i))