def simPheno(options):
print 'importing covariance matrix'
if options.cfile is None: options.cfile=options.bfile
XX = readCovarianceMatrixFile(options.cfile,readEig=False)['K']
print 'simulating phenotypes'
SP.random.seed(options.seed)
simulator = sim.CSimulator(bfile=options.bfile,XX=XX,P=options.nTraits)
Xr,region = simulator.getRegion(chrom_i=options.chrom,size=options.windowSize,min_nSNPs=options.nCausalR,pos_min=options.pos_min,pos_max=options.pos_max)
Y,info = genPhenoCube(simulator,Xr,vTotR=options.vTotR,nCausalR=options.nCausalR,pCommonR=options.pCommonR,vTotBg=options.vTotBg,pHidd=options.pHidden,pCommon=options.pCommon)
print 'exporting pheno file'
if options.pfile is not None:
outdir = os.path.split(options.pfile)[0]
if not os.path.exists(outdir):
os.makedirs(outdir)
else:
identifier = '_seed%d_nTraits%d_wndSize%d_vTotR%.2f_nCausalR%d_pCommonR%.2f_vTotBg%.2f_pHidden%.2f_pCommon%.2f'%(options.seed,options.nTraits,options.windowSize,options.vTotR,options.nCausalR,options.pCommonR,options.vTotBg,options.pHidden,options.pCommon)
options.pfile = os.path.split(options.bfile)[-1] + '%s'%identifier
pfile = options.pfile + '.phe'
rfile = options.pfile + '.phe.region'
SP.savetxt(pfile,Y)
SP.savetxt(rfile,region)
def analyze(options):
# load data
print 'import data'
if options.cfile is None:
cov = {'eval': None, 'evec': None}
warnings.warn(
'warning: cfile not specifed, a one variance compoenent model will be considered'
)
else:
cov = readCovarianceMatrixFile(options.cfile, readCov=False)
Y = readPhenoFile(options.pfile, idx=options.trait_idx)
null = readNullModelFile(options.nfile)
wnds = readWindowsFile(options.wfile)
F = None
if options.ffile:
F = readCovariatesFile(options.ffile)
#null['params_mean'] = SP.loadtxt(options.nfile + '.f0')
if F is not None: assert Y.shape[0] == F.shape[0], 'dimensions mismatch'
if options.i0 is None: options.i0 = 1
if options.i1 is None: options.i1 = wnds.shape[0]
# name of output file
if options.perm_i is not None:
res_dir = os.path.join(options.resdir, 'perm%d' % options.perm_i)
else:
res_dir = os.path.join(options.resdir, 'test')
if not os.path.exists(res_dir):
os.makedirs(res_dir)
n_digits = len(str(wnds.shape[0]))
fname = str(options.i0).zfill(n_digits)
fname += '_' + str(options.i1).zfill(n_digits) + '.res'
resfile = os.path.join(res_dir, fname)
# analysis
t0 = time.time()
scan(options.bfile, Y, cov, null, wnds, options.minSnps, options.i0,
options.i1, options.perm_i, resfile, F)
t1 = time.time()
print '... finished in %s seconds' % (t1 - t0)
def analyze(options):
# load data
print 'import data'
if options.cfile is None:
cov = {'eval':None,'evec':None}
warnings.warn('warning: cfile not specifed, a one variance compoenent model will be considered')
else:
cov = readCovarianceMatrixFile(options.cfile,readCov=False)
Y = readPhenoFile(options.pfile,idx=options.trait_idx)
null = readNullModelFile(options.nfile)
wnds = readWindowsFile(options.wfile)
F = None
if options.ffile:
F = readCovariatesFile(options.ffile)
#null['params_mean'] = SP.loadtxt(options.nfile + '.f0')
if F is not None: assert Y.shape[0]==F.shape[0], 'dimensions mismatch'
if options.i0 is None: options.i0 = 1
if options.i1 is None: options.i1 = wnds.shape[0]
# name of output file
if options.perm_i is not None:
res_dir = os.path.join(options.resdir,'perm%d'%options.perm_i)
else:
res_dir = os.path.join(options.resdir,'test')
if not os.path.exists(res_dir):
os.makedirs(res_dir)
n_digits = len(str(wnds.shape[0]))
fname = str(options.i0).zfill(n_digits)
fname+= '_'+str(options.i1).zfill(n_digits)+'.res'
resfile = os.path.join(res_dir,fname)
# analysis
t0 = time.time()
scan(options.bfile,Y,cov,null,wnds,options.minSnps,options.i0,options.i1,options.perm_i,resfile,F)
t1 = time.time()
print '... finished in %s seconds'%(t1-t0)
def preprocess(options):
assert options.bfile!=None, 'Please specify a bfile.'
""" computing the covariance matrix """
if options.compute_cov:
assert options.bfile!=None, 'Please specify a bfile.'
assert options.cfile is not None, 'Specify covariance matrix basename'
print 'Computing covariance matrix'
t0 = time.time()
computeCovarianceMatrix(options.plink_path,options.bfile,options.cfile,options.sim_type)
t1 = time.time()
print '... finished in %s seconds'%(t1-t0)
print 'Computing eigenvalue decomposition'
t0 = time.time()
eighCovarianceMatrix(options.cfile)
t1 = time.time()
print '... finished in %s seconds'%(t1-t0)
""" computing principal components """
if options.compute_PCs>0:
assert options.ffile is not None, 'Specify fix effects basename for saving PCs'
t0 = time.time()
computePCs(options.plink_path,options.compute_PCs,options.bfile,options.ffile)
t1 = time.time()
print '... finished in %s seconds'%(t1-t0)
""" fitting the null model """
if options.fit_null:
if options.nfile is None:
options.nfile = os.path.split(options.bfile)[-1]
warnings.warn('nfile not specifed, set to %s'%options.nfile)
print 'Fitting null model'
assert options.pfile is not None, 'phenotype file needs to be specified'
# read pheno
Y = readPhenoFile(options.pfile,idx=options.trait_idx)
# read covariance
if options.cfile is None:
cov = {'eval':None,'evec':None}
warnings.warn('cfile not specifed, a one variance compoenent model will be considered')
else:
cov = readCovarianceMatrixFile(options.cfile,readCov=False)
assert Y.shape[0]==cov['eval'].shape[0], 'dimension mismatch'
# read covariates
F = None
if options.ffile is not None:
F = readCovariatesFile(options.ffile)
assert Y.shape[0]==F.shape[0], 'dimensions mismatch'
t0 = time.time()
fit_null(Y,cov['eval'],cov['evec'],options.nfile, F)
t1 = time.time()
print '.. finished in %s seconds'%(t1-t0)
""" precomputing the windows """
if options.precompute_windows:
if options.wfile==None:
options.wfile = os.path.split(options.bfile)[-1] + '.%d'%options.window_size
warnings.warn('wfile not specifed, set to %s'%options.wfile)
print 'Precomputing windows'
t0 = time.time()
pos = readBimFile(options.bfile)
nWnds,nSnps=splitGeno(pos,size=options.window_size,out_file=options.wfile+'.wnd')
print 'Number of variants:',pos.shape[0]
print 'Number of windows:',nWnds
print 'Minimum number of snps:',nSnps.min()
print 'Maximum number of snps:',nSnps.max()
t1 = time.time()
print '.. finished in %s seconds'%(t1-t0)
# plot distribution of nSnps
if options.plot_windows:
print 'Plotting ditribution of number of SNPs'
plot_file = options.wfile+'.wnd.pdf'
plt = PL.subplot(1,1,1)
PL.hist(nSnps,30)
PL.xlabel('Number of SNPs')
PL.ylabel('Number of windows')
PL.savefig(plot_file)