def calc_h2(pheno, prev, eigen, keepArr, covar, numRemovePCs, lowtail):
pheno = leapUtils._fixup_pheno(pheno)
#Extract phenotype
if isinstance(pheno, dict): phe = pheno['vals']
else: phe = pheno
if (len(phe.shape) == 2):
if (phe.shape[1] == 1): phe = phe[:, 0]
else: raise Exception('More than one phenotype found')
if (keepArr is None): keepArr = np.ones(phe.shape[0], dtype=np.bool)
#Compute kinship matrix
XXT = eigen['XXT']
#Remove top PCs from kinship matrix
if (numRemovePCs > 0):
if (eigen is None): S, U = leapUtils.eigenDecompose(XXT)
else: S, U = eigen['arr_1'], eigen['arr_0']
print 'Removing the top', numRemovePCs, 'PCs from the kinship matrix'
XXT -= (U[:, -numRemovePCs:] * S[-numRemovePCs:]).dot(
U[:, -numRemovePCs:].T)
else:
U, S = None, None
#Determine if this is a case-control study
pheUnique = np.unique(phe)
if (pheUnique.shape[0] < 2):
raise Exception('Less than two different phenotypes observed')
isCaseControl = (pheUnique.shape[0] == 2)
if isCaseControl:
print 'Computing h2 for a binary phenotype'
pheMean = phe.mean()
phe[phe <= pheMean] = 0
phe[phe > pheMean] = 1
if (numRemovePCs > 0 or covar is not None):
probs, thresholds = calcLiabThreholds(U, S, keepArr, phe,
numRemovePCs, prev, covar)
h2 = calcH2Binary(XXT, phe, probs, thresholds, keepArr, prev)
else:
h2 = calcH2Binary(XXT, phe, None, None, keepArr, prev)
else:
if (covar is not None):
raise Exception(
'Covariates with a continuous phenotype are currently not supported'
)
print 'Computing h2 for a continuous phenotype'
if (not lowtail): h2 = calcH2Continuous(XXT, phe, keepArr, prev)
else: h2 = calcH2Continuous_twotails(XXT, phe, keepArr, prev)
if (h2 <= 0): raise Exception("Negative heritability found. Exitting...")
if (np.isnan(h2)):
raise Exception(
"Invalid heritability estimate. Please double-check your input for any errors."
)
print 'h2: %0.6f' % h2
return h2
def calc_h2(pheno, prev, eigen, keepArr, covar, numRemovePCs, h2coeff, lowtail):
pheno = leapUtils._fixup_pheno(pheno)
#Extract phenotype
if isinstance(pheno, dict): phe = pheno['vals']
else: phe = pheno
if (len(phe.shape)==2):
if (phe.shape[1]==1): phe=phe[:,0]
else: raise Exception('More than one phenotype found')
if (keepArr is None): keepArr = np.ones(phe.shape[0], dtype=np.bool)
#Compute kinship matrix
XXT = eigen['XXT']
#Remove top PCs from kinship matrix
if (numRemovePCs > 0):
if (eigen is None): S,U = leapUtils.eigenDecompose(XXT)
else: S, U = eigen['arr_1'], eigen['arr_0']
print 'Removing the top', numRemovePCs, 'PCs from the kinship matrix'
XXT -= (U[:, -numRemovePCs:]*S[-numRemovePCs:]).dot(U[:, -numRemovePCs:].T)
else:
U, S = None, None
#Determine if this is a case-control study
pheUnique = np.unique(phe)
if (pheUnique.shape[0] < 2): raise Exception('Less than two different phenotypes observed')
isCaseControl = (pheUnique.shape[0] == 2)
if isCaseControl:
print 'Computing h2 for a binary phenotype'
pheMean = phe.mean()
phe[phe <= pheMean] = 0
phe[phe > pheMean] = 1
if (numRemovePCs > 0 or covar is not None):
probs, thresholds = calcLiabThreholds(U, S, keepArr, phe, numRemovePCs, prev, covar)
h2 = calcH2Binary(XXT, phe, probs, thresholds, keepArr, prev, h2coeff)
else: h2 = calcH2Binary(XXT, phe, None, None, keepArr, prev, h2coeff)
else:
if (covar is not None): raise Exception('Covariates with a continuous phenotype are currently not supported')
print 'Computing h2 for a continuous phenotype'
if (not lowtail): h2 = calcH2Continuous(XXT, phe, keepArr, prev, h2coeff)
else: h2 = calcH2Continuous_twotails(XXT, phe, keepArr, prev, h2coeff)
if (h2 <= 0): raise Exception("Negative heritability found. Exitting...")
if (np.isnan(h2)): raise Exception("Invalid heritability estimate. Please double-check your input for any errors.")
print 'h2: %0.6f'%h2
return h2
def eigenDecompose(bed, outFile=None): bed = leapUtils._fixupBed(bed) #Compute kinship matrix t0 = time.time() print 'Computing kinship matrix...' XXT = leapUtils.symmetrize(blas.dsyrk(1.0, bed.val, lower=1)) / bed.val.shape[1] print 'Done in %0.2f'%(time.time()-t0), 'seconds' #Compute eigendecomposition S,U = leapUtils.eigenDecompose(XXT) if (outFile is not None): np.savez_compressed(outFile, arr_0=U, arr_1=S, XXT=XXT) eigen = dict([]) eigen['XXT'] = XXT eigen['arr_0'] = U eigen['arr_1'] = S return eigen
def eigenDecompose(bed, outFile):
bed = leapUtils._fixupBed(bed)
# Compute kinship matrix
t0 = time.time()
print "Computing kinship matrix..."
XXT = leapUtils.symmetrize(blas.dsyrk(1.0, bed.val, lower=1)) / bed.val.shape[1]
print "Done in %0.2f" % (time.time() - t0), "seconds"
# Compute eigendecomposition
S, U = leapUtils.eigenDecompose(XXT)
if outFile is not None:
np.savez_compressed(outFile, arr_0=U, arr_1=S, XXT=XXT)
eigen = dict([])
eigen["XXT"] = XXT
eigen["arr_0"] = U
eigen["arr_1"] = S
return eigen
def eigenDecompose(bed, outFile=None):
bed = leapUtils._fixupBed(bed)
#Compute kinship matrix
t0 = time.time()
print 'Computing kinship matrix...'
XXT = leapUtils.symmetrize(blas.dsyrk(1.0, bed.val,
lower=1)) / bed.val.shape[1]
print 'Done in %0.2f' % (time.time() - t0), 'seconds'
#Compute eigendecomposition
S, U = leapUtils.eigenDecompose(XXT)
if (outFile is not None):
np.savez_compressed(outFile, arr_0=U, arr_1=S, XXT=XXT)
eigen = dict([])
eigen['XXT'] = XXT
eigen['arr_0'] = U
eigen['arr_1'] = S
return eigen
