def lmem(args, start, stop, output):
print "Processing SNPs", start, "-", stop, "..."
print "Reading phenotypes and RRM..."
Y = np.genfromtxt(iolib.ropen(args.phenotype),
missing_values="NA",
dtype=np.float,
skip_header=args.header)
if len(Y.shape) == 1:
tmp = np.empty((len(Y), 1), np.float)
tmp[:, 0] = Y
Y = tmp
nsample, npheno = Y.shape
genfile = iolib.genReader(args.genfile, nsample, args.linebuffer, start,
stop)
outf = csv.writer(iolib.wopen(output), delimiter="\t")
K = np.genfromtxt(iolib.ropen(args.kinship), dtype=np.float)
print "K.shape =", K.shape
print "Y.shape =", Y.shape
print nsample, "samples and", npheno, "phenotypes"
missing = []
null_model = []
print "Calculating rotation matrices..."
for mpheno in range(npheno):
missing.append(np.logical_not(np.isnan(Y[:, mpheno])))
ii = missing[mpheno]
null_model.append(lmem_func.get_delta(K[ii][:, ii], Y[ii, mpheno]))
# Ut,denom,beta_null,sigma_g_null,loglik_null,delta_null = get_delta(K[ii][:,ii],Y[ii,mpheno]))
print "done"
print "Fitting LMMs..."
time0 = time.time()
processed = 0
for rsid1, rsid2, pos, G in genfile:
output = []
nsnp = G.shape[1]
for mpheno in range(npheno):
ii = missing[mpheno]
output.append(
lmem_func.fitlmm(null_model[mpheno]['offset'],
null_model[mpheno]['Ut'],
null_model[mpheno]['denom'], G[ii],
null_model[mpheno]['uty']))
for i in range(nsnp):
outf.writerow([rsid1[i], pos[i], rsid2[i]] + sum([[
null_model[mpheno]['beta_null'], null_model[mpheno]
['sigma_g_null'], null_model[mpheno]['loglik_null']
] + output[mpheno][0][i].tolist() + [
output[mpheno][1][i], output[mpheno][2][i], output[mpheno][3][i]
] for mpheno in range(npheno)], []))
processed += nsnp
print processed, "loci processed"
print "Took", time.time() - time0, "seconds"
def lmem(args,start,stop,output):
print "Processing SNPs",start,"-",stop,"..."
print "Reading phenotypes and RRM..."
Y = np.genfromtxt(iolib.ropen(args.phenotype),missing_values="NA",dtype=np.float,skip_header=args.header)
if len(Y.shape)==1:
tmp = np.empty((len(Y),1),np.float)
tmp[:,0] = Y
Y = tmp
nsample,npheno = Y.shape
genfile = iolib.genReader(args.genfile,nsample,args.linebuffer,start,stop)
outf = csv.writer(iolib.wopen(output),delimiter="\t")
K = np.genfromtxt(iolib.ropen(args.kinship),dtype=np.float)
print "K.shape =",K.shape
print "Y.shape =",Y.shape
print nsample,"samples and",npheno,"phenotypes"
missing = []
null_model = []
print "Calculating rotation matrices..."
for mpheno in range(npheno):
missing.append(np.logical_not(np.isnan(Y[:,mpheno])))
ii = missing[mpheno]
null_model.append(lmem_func.get_delta(K[ii][:,ii],Y[ii,mpheno]))
# Ut,denom,beta_null,sigma_g_null,loglik_null,delta_null = get_delta(K[ii][:,ii],Y[ii,mpheno]))
print "done"
print "Fitting LMMs..."
time0 = time.time()
processed = 0
for rsid1,rsid2,pos,G in genfile:
output = []
nsnp = G.shape[1]
for mpheno in range(npheno):
ii = missing[mpheno]
output.append(lmem_func.fitlmm(null_model[mpheno]['offset'],null_model[mpheno]['Ut'],null_model[mpheno]['denom'],G[ii],null_model[mpheno]['uty']))
for i in range(nsnp):
outf.writerow([rsid1[i],pos[i],rsid2[i]]
+sum([[null_model[mpheno]['beta_null'],null_model[mpheno]['sigma_g_null'],null_model[mpheno]['loglik_null']]
+output[mpheno][0][i].tolist()+[output[mpheno][1][i],output[mpheno][2][i],output[mpheno][3][i]] for mpheno in range(npheno)],[]))
processed+=nsnp
print processed,"loci processed"
print "Took",time.time() - time0,"seconds"
#!/usr/bin/python -O
import sys,time,os,glob,pickle,cPickle, csv,gzip,numpy as np,resource,argparse,string
import numpy as np
import iolib
parser = argparse.ArgumentParser(description='calculates the RRM on plink binary file set')
parser.add_argument('plinkfile', metavar='plinkfile', type=str, help='a binary plink file set')
parser.add_argument('-snps', metavar='snps', type=str,default='', help='list of snp ids to use for rrm calculation')
parser.add_argument('-out', metavar='out', type=str,default='', help='outfile')
parser.add_argument('--ibs', metavar='out', type=bool,default=False, help='use kinship coefficient rather than realised relationship')
args = parser.parse_args()
if args.out=='': outfile = csv.writer(iolib.wopen(args.plinkfile+".rrm.gz"),delimiter="\t")
else: outfile = csv.writer(iolib.wopen(args.out+".rrm.gz"),delimiter="\t")
if args.snps!='':
snps = iolib.scan(args.snps)
print "Calculating RRM from subset of",len(snps),"SNPs"
else: snps = None
infile = iolib.plinkReader(args.plinkfile,snps=snps)
n = infile.nsample
rrm = np.zeros((n,n),np.float)
rrm_diag = np.zeros(n,np.float)
print "Calculating RRM..."
processed += nsnp
print processed, "loci processed"
print "Took", time.time() - time0, "seconds"
if __name__ == '__main__':
# logger = mp.log_to_stderr()
# logger.setLevel(logging.INFO)
assert args.header >= 0
iolib.checkfile(args.kinship)
iolib.checkfile(args.genfile)
if args.output[-3:] == ".gz": args.output = args.output[:-3]
outf = csv.writer(iolib.wopen(args.output + ".gz"), delimiter="\t")
nl = iolib.nlines(iolib.ropen(args.genfile))
print nl, "SNPs"
neach = int(math.ceil(nl / args.nprocess))
chunks = range(0, nl, neach) + [nl]
print chunks
pool = mp.Pool(processes=args.nprocess)
Y = np.genfromtxt(iolib.ropen(args.phenotype),
missing_values="NA",
dtype=np.float,
skip_header=args.header)
if len(Y.shape) == 1:
processed+=nsnp
print processed,"loci processed"
print "Took",time.time() - time0,"seconds"
if __name__ == '__main__':
# logger = mp.log_to_stderr()
# logger.setLevel(logging.INFO)
assert args.header>=0
iolib.checkfile(args.kinship)
iolib.checkfile(args.genfile)
if args.output[-3:]==".gz": args.output = args.output[:-3]
outf = csv.writer(iolib.wopen(args.output+".gz"),delimiter="\t")
nl = iolib.nlines(iolib.ropen(args.genfile))
print nl,"SNPs"
neach = int(math.ceil(nl/args.nprocess))
chunks = range(0,nl,neach) + [nl]
print chunks
pool = mp.Pool(processes=args.nprocess)
Y = np.genfromtxt(iolib.ropen(args.phenotype),missing_values="NA",dtype=np.float,skip_header=args.header)
if len(Y.shape)==1:
nsample = Y.shape[0]
npheno = 1
else: