def _test_():
singleton_snps = genotypes.simulate_k_tons(n=500, m=1000)
doubleton_snps = genotypes.simulate_k_tons(k=2, n=500, m=1000)
common_snps = genotypes.simulate_common_genotypes(500, 1000)
snps = sp.vstack([common_snps, singleton_snps, doubleton_snps])
print snps
snps = snps.T
snps = (snps - sp.mean(snps, 0)) / sp.std(snps, 0)
snps = snps.T
print snps, snps.shape
file_prefix = os.environ['HOME'] + '/tmp/test'
phen_list = phenotypes.simulate_traits_w_snps_to_hdf5(snps, hdf5_file_prefix=file_prefix,
num_traits=30, p=0.1)
singletons_thres = []
doubletons_thres = []
common_thres = []
for i, y in enumerate(phen_list['phenotypes']):
K = kinship.calc_ibd_kinship(snps)
K = kinship.scale_k(K)
lmm = lm.LinearMixedModel(y)
lmm.add_random_effect(K)
r1 = lmm.get_REML()
print 'pseudo_heritability:', r1['pseudo_heritability']
ex_res = lm.emmax(snps, y, K)
plt.figure()
plt.hist(y, 50)
plt.savefig('%s_%d_phen.png' % (file_prefix, i))
plt.clf()
agr.plot_simple_qqplots_pvals('%s_%d' % (file_prefix, i),
[ex_res['ps'][:1000], ex_res['ps'][1000:2000], ex_res['ps'][2000:]],
result_labels=['Common SNPs', 'Singletons', 'Doubletons'],
line_colors=['b', 'r', 'y'],
num_dots=200, max_neg_log_val=3)
# Cholesky permutations..
res = lm.emmax_perm_test(singleton_snps, y, K, num_perm=1000)
print 1.0 / (20 * 1000.0), res['threshold_05']
singletons_thres.append(res['threshold_05'][0])
res = lm.emmax_perm_test(doubleton_snps, y, K, num_perm=1000)
print 1.0 / (20 * 1000.0), res['threshold_05']
doubletons_thres.append(res['threshold_05'][0])
res = lm.emmax_perm_test(common_snps, y, K, num_perm=1000)
print 1.0 / (20 * 1000.0), res['threshold_05']
common_thres.append(res['threshold_05'][0])
#ATT permutations (Implement)
#PC permutations (Implement)
print sp.mean(singletons_thres), sp.std(singletons_thres)
print sp.mean(doubletons_thres), sp.std(doubletons_thres)
print sp.mean(common_thres), sp.std(common_thres)
def _test_scz_():
# Load Schizophrenia data
singleton_snps = genotypes.simulate_k_tons(n=500, m=1000)
doubleton_snps = genotypes.simulate_k_tons(k=2, n=500, m=1000)
common_snps = genotypes.simulate_common_genotypes(500, 1000)
snps = sp.vstack([common_snps, singleton_snps, doubleton_snps])
test_snps = sp.vstack([singleton_snps, doubleton_snps])
print snps
phen_list = phenotypes.simulate_traits(
snps, hdf5_file_prefix='/home/bv25/tmp/test', num_traits=30, p=1.0)
singletons_thres = []
doubletons_thres = []
common_thres = []
for i, y in enumerate(phen_list):
K = kinship.calc_ibd_kinship(snps)
K = kinship.scale_k(K)
lmm = lm.LinearMixedModel(y)
lmm.add_random_effect(K)
r1 = lmm.get_REML()
print 'pseudo_heritability:', r1['pseudo_heritability']
ex_res = lm.emmax(snps, y, K)
plt.figure()
plt.hist(y, 50)
plt.savefig('/home/bv25/tmp/test_%d_phen.png' % i)
plt.clf()
agr.plot_simple_qqplots_pvals('/home/bv25/tmp/test_%d' % i, [
ex_res['ps'][:1000], ex_res['ps'][1000:2000], ex_res['ps'][2000:]
],
result_labels=[
'Common SNPs', 'Singletons',
'Doubletons'
],
line_colors=['b', 'r', 'y'],
num_dots=200,
max_neg_log_val=3)
# Now permutations..
res = lm.emmax_perm_test(singleton_snps, y, K, num_perm=1000)
print 1.0 / (20 * 1000.0), res['threshold_05']
singletons_thres.append(res['threshold_05'][0])
res = lm.emmax_perm_test(doubleton_snps, y, K, num_perm=1000)
print 1.0 / (20 * 1000.0), res['threshold_05']
doubletons_thres.append(res['threshold_05'][0])
res = lm.emmax_perm_test(common_snps, y, K, num_perm=1000)
print 1.0 / (20 * 1000.0), res['threshold_05']
common_thres.append(res['threshold_05'][0])
print sp.mean(singletons_thres), sp.std(singletons_thres)
print sp.mean(doubletons_thres), sp.std(doubletons_thres)
print sp.mean(common_thres), sp.std(common_thres)
def _test_scz_():
# Load Schizophrenia data
singleton_snps = genotypes.simulate_k_tons(n=500, m=1000)
doubleton_snps = genotypes.simulate_k_tons(k=2, n=500, m=1000)
common_snps = genotypes.simulate_common_genotypes(500, 1000)
snps = sp.vstack([common_snps, singleton_snps, doubleton_snps])
test_snps = sp.vstack([singleton_snps, doubleton_snps])
print snps
phen_list = phenotypes.simulate_traits(snps, hdf5_file_prefix='/home/bv25/tmp/test', num_traits=30, p=1.0)
singletons_thres = []
doubletons_thres = []
common_thres = []
for i, y in enumerate(phen_list):
K = kinship.calc_ibd_kinship(snps)
K = kinship.scale_k(K)
lmm = lm.LinearMixedModel(y)
lmm.add_random_effect(K)
r1 = lmm.get_REML()
print 'pseudo_heritability:', r1['pseudo_heritability']
ex_res = lm.emmax(snps, y, K)
plt.figure()
plt.hist(y, 50)
plt.savefig('/home/bv25/tmp/test_%d_phen.png' % i)
plt.clf()
agr.plot_simple_qqplots_pvals('/home/bv25/tmp/test_%d' % i,
[ex_res['ps'][:1000], ex_res['ps'][1000:2000], ex_res['ps'][2000:]],
result_labels=['Common SNPs', 'Singletons', 'Doubletons'],
line_colors=['b', 'r', 'y'],
num_dots=200, max_neg_log_val=3)
# Now permutations..
res = lm.emmax_perm_test(singleton_snps, y, K, num_perm=1000)
print 1.0 / (20 * 1000.0), res['threshold_05']
singletons_thres.append(res['threshold_05'][0])
res = lm.emmax_perm_test(doubleton_snps, y, K, num_perm=1000)
print 1.0 / (20 * 1000.0), res['threshold_05']
doubletons_thres.append(res['threshold_05'][0])
res = lm.emmax_perm_test(common_snps, y, K, num_perm=1000)
print 1.0 / (20 * 1000.0), res['threshold_05']
common_thres.append(res['threshold_05'][0])
print sp.mean(singletons_thres), sp.std(singletons_thres)
print sp.mean(doubletons_thres), sp.std(doubletons_thres)
print sp.mean(common_thres), sp.std(common_thres)
def test_null_fig2(n=100, num_traits=1000, corr=0.8, beta_c=0, beta_y=0, alpha_thres=0.05):
"""
# 1. Generate test genotypes.
# 2. Generate GE1 and GE2, w correlation.
# 3. Construct phenotypes.
# 4. Check direct association signal.
# 5. Check covariate adjusted association signal.
"""
snps = genotypes.simulate_common_genotypes(n=n, m=num_traits)
# Normalize genotypes
snps = snps.T
snps = (snps - sp.mean(snps, 0)) / sp.std(snps, 0)
snps = snps.T
V = sp.kron(sp.array([[1, corr], [corr, 1]]), sp.eye(n))
# print V
GE = sp.random.multivariate_normal(sp.zeros(n * 2), cov=V, size=num_traits)
# print GE.shape
GE1 = GE[:, :n]
GE1_norm = GE1.T
GE1_norm = (GE1_norm - sp.mean(GE1_norm, 0)) / sp.std(GE1_norm, 0)
GE1_norm = GE1_norm.T
GE2 = GE[:, n:]
GE2_norm = GE2.T
GE2_norm = (GE2_norm - sp.mean(GE2_norm, 0)) / sp.std(GE2_norm, 0)
GE2_norm = GE2_norm.T
gs = snps
C = GE1_norm + beta_c * gs
# print C[0], GE1_norm[0],gs[0]
# print sp.var(C[0]), sp.var(GE1_norm[0]), sp.var(gs[0])
# print C.shape, GE1.shape, gs.shape
Y = GE2_norm + beta_y * gs
# #Normalize Y's and C's
# C_norm = C.T
# C_norm_factor = sp.std(C_norm, 0)
# C_norm = (C_norm - sp.mean(C_norm, 0))/ C_norm_factor
# C = C_norm.T
#
# Y_norm = Y.T
# Y_norm = (Y_norm - sp.mean(Y_norm, 0)) / sp.std(Y_norm, 0)
# Y = Y_norm.T
adj_betas_Y = sp.empty((num_traits, 2))
for i in range(num_traits):
X = sp.vstack([snps[i], C[i]])
XX = sp.mat(sp.dot(X, X.T))
XX_inv = linalg.inv(XX)
adj_betas_Y[i] = sp.dot(XX_inv, sp.dot(X, Y[i].T).T)
# print adj_betas_Y.shape
print "Adj association on Y:"
print sp.mean(adj_betas_Y, 0)
# print adj_betas_Y
adj_betas_Y_genot = adj_betas_Y[:, 0]
res0 = Y - (C.T * adj_betas_Y[:, 1]).T
rss0 = sp.sum(res0 ** 2, 1)
res1 = Y - (snps.T * adj_betas_Y_genot).T - (C.T * adj_betas_Y[:, 1]).T
# print res.shape
rss1 = sp.sum(res1 ** 2, 1)
f_stats = (rss0 - rss1) * (n - 2) / rss1
adj_Y_p_vals = stats.f.sf(f_stats, 1, n - 2)
mean_adj_pvals = sp.mean(adj_Y_p_vals)
print "Mean Adjust Y p-value:", mean_adj_pvals
if beta_y == 0:
adj_Y_fdr = sp.sum(adj_Y_p_vals < alpha_thres) / float(num_traits)
adj_Y_power = 0
else:
adj_Y_fdr = 0
adj_Y_power = sp.sum(adj_Y_p_vals < alpha_thres) / float(num_traits)
print "Adj Y power:", adj_Y_power
print "Adj Y FDR:", adj_Y_fdr
adj_Y_bias = sp.mean(beta_y - adj_betas_Y_genot)
ret_dict = {
"mean_adj_pvals": mean_adj_pvals,
"adj_Y_power": adj_Y_power,
"adj_Y_fdr": adj_Y_fdr,
"adj_Y_bias": adj_Y_bias,
}
return ret_dict
def test_null(n=100, num_traits=1000, corr=0.8, beta_c=0, beta_y=0, alpha_thres=0.05):
"""
# 1. Generate test genotypes.
# 2. Generate GE1 and GE2, w correlation.
# 3. Construct phenotypes.
# 4. Check direct association signal.
# 5. Check covariate adjusted association signal.
"""
snps = genotypes.simulate_common_genotypes(n=n, m=num_traits)
# Normalize genotypes
snps = snps.T
snps = (snps - sp.mean(snps, 0)) / sp.std(snps, 0)
snps = snps.T
V = sp.kron(sp.array([[1, corr], [corr, 1]]), sp.eye(n))
# print V
GE = sp.random.multivariate_normal(sp.zeros(n * 2), cov=V, size=num_traits)
# print GE.shape
GE1 = GE[:, :n]
GE1_norm = GE1.T
GE1_norm = (GE1_norm - sp.mean(GE1_norm, 0)) / sp.std(GE1_norm, 0)
GE1_norm = GE1_norm.T
GE2 = GE[:, n:]
GE2_norm = GE2.T
GE2_norm = (GE2_norm - sp.mean(GE2_norm, 0)) / sp.std(GE2_norm, 0)
GE2_norm = GE2_norm.T
gs = snps
C = GE1_norm + beta_c * gs
# print C[0], GE1_norm[0],gs[0]
# print sp.var(C[0]), sp.var(GE1_norm[0]), sp.var(gs[0])
# print C.shape, GE1.shape, gs.shape
Y = GE2_norm + beta_y * gs
# Normalize Y's and C's
C_norm = C.T
C_norm_factor = sp.std(C_norm, 0)
C_norm = (C_norm - sp.mean(C_norm, 0)) / C_norm_factor
C_norm = C_norm.T
Y_norm = Y.T
Y_norm = (Y_norm - sp.mean(Y_norm, 0)) / sp.std(Y_norm, 0)
Y_norm = Y_norm.T
est_corrs = sp.zeros(num_traits)
for i in range(num_traits):
est_corrs[i] = sp.corrcoef(C_norm[i], Y_norm[i])[0, 1]
print sp.mean(est_corrs)
direct_betas_Y = sp.empty(num_traits)
for i in range(num_traits):
direct_betas_Y[i] = sp.dot(snps[i], Y[i]) / n
print "Direct association on Y:"
# print sp.mean(direct_betas_Y), sp.var(direct_betas_Y)
rss0 = sp.sum(Y ** 2, 1)
res = Y - (snps.T * direct_betas_Y).T
# print res.shape
rss1 = sp.sum(res ** 2, 1)
f_stats = (rss0 - rss1) * (n - 1) / rss1
marg_Y_p_vals = stats.f.sf(f_stats, 1, n - 1)
mean_marg_Y_pvals = sp.mean(marg_Y_p_vals)
print "Mean marginal Y p-value:", mean_marg_Y_pvals
if beta_y == 0:
marg_Y_fdr = sp.sum(marg_Y_p_vals < alpha_thres) / float(num_traits)
marg_Y_power = 0
else:
marg_Y_fdr = 0
marg_Y_power = sp.sum(marg_Y_p_vals < alpha_thres) / float(num_traits)
print "Marginal Y power:", marg_Y_power
print "Marginal Y FDR:", marg_Y_fdr
direct_betas_C = sp.empty(num_traits)
for i in range(num_traits):
direct_betas_C[i] = sp.dot(snps[i], C[i]) / n
# print len(direct_betas_C)
print "Direct association on C:"
print sp.mean(direct_betas_C), sp.var(direct_betas_C)
rss0 = sp.sum(C ** 2, 1)
C_SNP_resid = C - (snps.T * direct_betas_C).T
# print res.shape
rss1 = sp.sum(C_SNP_resid ** 2, 1)
f_stats = (rss0 - rss1) * (n - 1) / rss1
marg_C_p_vals = stats.f.sf(f_stats, 1, n - 1)
mean_marg_C_pvals = sp.mean(marg_C_p_vals)
print "Mean marginal C p-value:", mean_marg_C_pvals
if beta_c == 0:
marg_C_fdr = sp.sum(marg_C_p_vals < alpha_thres) / float(num_traits)
marg_C_power = 0
else:
marg_C_fdr = 0
marg_C_power = sp.sum(marg_C_p_vals < alpha_thres) / float(num_traits)
print "Marginal C power:", marg_C_power
print "Marginal C FDR:", marg_C_fdr
adj_betas_Y = sp.empty((num_traits, 2))
for i in range(num_traits):
X = sp.vstack([snps[i], C[i]])
XX = sp.mat(sp.dot(X, X.T))
XX_inv = linalg.inv(XX)
adj_betas_Y[i] = sp.dot(XX_inv, sp.dot(X, Y[i].T).T)
# print adj_betas_Y.shape
print "Adj association on Y:"
print sp.mean(adj_betas_Y, 0)
# print adj_betas_Y
adj_betas_Y_genot = adj_betas_Y[:, 0]
res0 = Y - (C.T * adj_betas_Y[:, 1]).T
rss0 = sp.sum(res0 ** 2, 1)
res1 = Y - (snps.T * adj_betas_Y_genot).T - (C.T * adj_betas_Y[:, 1]).T
# print res.shape
rss1 = sp.sum(res1 ** 2, 1)
f_stats = (rss0 - rss1) * (n - 2) / rss1
adj_Y_p_vals = stats.f.sf(f_stats, 1, n - 2)
mean_adj_pvals = sp.mean(adj_Y_p_vals)
print "Mean Adjust Y p-value:", mean_adj_pvals
if beta_y == 0:
adj_Y_fdr = sp.sum(adj_Y_p_vals < alpha_thres) / float(num_traits)
adj_Y_power = 0
else:
adj_Y_fdr = 0
adj_Y_power = sp.sum(adj_Y_p_vals < alpha_thres) / float(num_traits)
print "Adj Y power:", adj_Y_power
print "Adj Y FDR:", adj_Y_fdr
adj_Y_bias = sp.mean(beta_y - adj_betas_Y_genot)
# Test for detecting a false association...
z_stats = sp.sqrt(n) * (adj_betas_Y_genot - direct_betas_Y) / est_corrs
bias_p_vals = stats.f.sf(z_stats ** 2, 1, n - 1)
mean_bias_p_val = sp.mean(bias_p_vals)
if beta_c == 0:
bias_test_fdr = sp.sum(bias_p_vals < alpha_thres) / float(num_traits)
bias_test_power = 0
else:
bias_test_fdr = 0
bias_test_power = sp.sum(bias_p_vals < alpha_thres) / float(num_traits)
print "Adj bias-test power:", bias_test_power
print "Adj bias-test FDR:", bias_test_fdr
# The conservative approach...
ok_filter = sp.sign(direct_betas_Y) == sp.sign(adj_betas_Y_genot)
ok_filter[ok_filter] = direct_betas_Y[ok_filter] ** 2 > adj_betas_Y_genot[ok_filter] ** 2
print sp.sum(ok_filter)
print len(ok_filter)
ok_pvals = sp.copy(marg_Y_p_vals)
ok_pvals[ok_filter] = adj_Y_p_vals[ok_filter]
increased_power_frac = sp.sum(ok_pvals < marg_Y_p_vals) / float(num_traits)
# Try regressing the genetic effect on C out of C and then perform the regression:
# Frequentists approach
adj_C_resid_betas_Y = sp.empty((num_traits, 2))
for i in range(num_traits):
X = sp.vstack([snps[i], C_SNP_resid[i]])
XX = sp.mat(sp.dot(X, X.T))
XX_inv = linalg.inv(XX)
adj_C_resid_betas_Y[i] = sp.dot(XX_inv, sp.dot(X, Y[i].T).T)
# print adj_betas_Y.shape
print "Adj association on Y C-g:"
print sp.mean(adj_betas_Y, 0)
# print adj_betas_Y
adj_C_resid_betas_Y_genot = adj_C_resid_betas_Y[:, 0]
adj_C_resid_betas_Y_C = adj_C_resid_betas_Y[:, 1]
C_effect = (C_SNP_resid.T * adj_C_resid_betas_Y_C).T
res0 = Y - C_effect
rss0 = sp.sum(res0 ** 2, 1)
res1 = Y - (snps.T * adj_C_resid_betas_Y_genot).T - C_effect
# print res.shape
rss1 = sp.sum(res1 ** 2, 1)
f_stats = (rss0 - rss1) * (n - 2) / rss1
adj_C_resid_Y_p_vals = stats.f.sf(f_stats, 1, n - 2)
mean_adj_C_resid_Y_p_val = sp.mean(adj_C_resid_Y_p_vals)
print "Mean Adjust Y (C-g) p-value :", mean_adj_C_resid_Y_p_val
print "Mean Adjust Y (C-g) bias:", sp.mean(beta_y - adj_C_resid_betas_Y_genot)
if beta_y == 0:
adj_C_resid_Y_fdr = sp.sum(adj_C_resid_Y_p_vals < alpha_thres) / float(num_traits)
adj_C_resid_Y_power = 0
else:
adj_C_resid_Y_fdr = 0
adj_C_resid_Y_power = sp.sum(adj_C_resid_Y_p_vals < alpha_thres) / float(num_traits)
print "Adj Y (C-g) power:", adj_C_resid_Y_power
print "Adj Y (C-g) FDR:", adj_C_resid_Y_fdr
# Bayesian approach
# 1. Calculate Bayes C_snp_resid.
p = 1 # Fraction of causal markers
beta_c2 = beta_c ** 2
h2 = (beta_c2) / (1 + beta_c2)
if beta_c == 0:
post_mean_direct_betas_C = 0
else:
post_mean_direct_betas_C = (1 / (1 + 1 / (n * h2))) * direct_betas_C
bayes_C_SNP_resid = C - (snps.T * post_mean_direct_betas_C).T
# 2. Use to calculate the statistic, the usual way.
bayes_adj_C_resid_betas_Y = sp.empty((num_traits, 2))
for i in range(num_traits):
X = sp.vstack([snps[i], bayes_C_SNP_resid[i]])
XX = sp.mat(sp.dot(X, X.T))
XX_inv = linalg.inv(XX)
bayes_adj_C_resid_betas_Y[i] = sp.dot(XX_inv, sp.dot(X, Y[i].T).T)
# print adj_betas_Y.shape
print "Adj association on Y:"
print sp.mean(adj_betas_Y, 0)
# print adj_betas_Y
bayes_adj_C_resid_betas_Y_genot = bayes_adj_C_resid_betas_Y[:, 0]
bayes_adj_C_resid_betas_Y_C = bayes_adj_C_resid_betas_Y[:, 1]
bayes_C_effect = (bayes_C_SNP_resid.T * bayes_adj_C_resid_betas_Y_C).T
res0 = Y - bayes_C_effect
rss0 = sp.sum(res0 ** 2, 1)
res1 = Y - (snps.T * bayes_adj_C_resid_betas_Y_genot).T - bayes_C_effect
# print res.shape
rss1 = sp.sum(res1 ** 2, 1)
f_stats = (rss0 - rss1) * (n - 2) / rss1
bayes_adj_C_resid_Y_p_vals = stats.f.sf(f_stats, 1, n - 2)
mean_bayes_adj_C_resid_Y_p_val = sp.mean(bayes_adj_C_resid_Y_p_vals)
print "Bayesian mean adjust Y (C-g) p-value :", mean_bayes_adj_C_resid_Y_p_val
print "Bayesian mean adjust Y (C-g) bias:", sp.mean(beta_y - adj_C_resid_betas_Y_genot)
if beta_y == 0:
bayes_adj_C_resid_Y_fdr = sp.sum(bayes_adj_C_resid_Y_p_vals < alpha_thres) / float(num_traits)
bayes_adj_C_resid_Y_power = 0
else:
bayes_adj_C_resid_Y_fdr = 0
bayes_adj_C_resid_Y_power = sp.sum(bayes_adj_C_resid_Y_p_vals < alpha_thres) / float(num_traits)
print "Bayesian adj Y (C-g) power:", bayes_adj_C_resid_Y_power
print "Bayesian adj Y (C-g) FDR:", bayes_adj_C_resid_Y_fdr
mean_ok_pvals = sp.mean(ok_pvals)
print "Mean OK p-value:", mean_ok_pvals
print "Mean bias p-value:", mean_bias_p_val
print "Increased power frac.:", increased_power_frac
ret_dict = {
"mean_ok_pvals": mean_ok_pvals,
"mean_adj_pvals": mean_adj_pvals,
"mean_marg_C_pvals": mean_marg_C_pvals,
"mean_marg_Y_pvals": mean_marg_Y_pvals,
"increased_power_frac": increased_power_frac,
"mean_bias_p_vals": mean_bias_p_val,
"marg_Y_power": marg_Y_power,
"marg_Y_fdr": marg_Y_fdr,
"marg_C_power": marg_C_power,
"marg_C_fdr": marg_C_fdr,
"adj_Y_power": adj_Y_power,
"adj_Y_fdr": adj_Y_fdr,
"bias_test_power": bias_test_power,
"bias_test_fdr": bias_test_fdr,
"adj_C_resid_Y_power": adj_C_resid_Y_power,
"adj_C_resid_Y_fdr": adj_C_resid_Y_fdr,
"bayes_adj_C_resid_Y_power": bayes_adj_C_resid_Y_power,
"bayes_adj_C_resid_Y_fdr": bayes_adj_C_resid_Y_fdr,
"adj_Y_bias": adj_Y_bias,
}
return ret_dict
def _test_():
singleton_snps = genotypes.simulate_k_tons(n=500, m=1000)
doubleton_snps = genotypes.simulate_k_tons(k=2, n=500, m=1000)
common_snps = genotypes.simulate_common_genotypes(500, 1000)
snps = sp.vstack([common_snps, singleton_snps, doubleton_snps])
print snps
snps = snps.T
snps = (snps - sp.mean(snps, 0)) / sp.std(snps, 0)
snps = snps.T
print snps, snps.shape
file_prefix = os.environ['HOME'] + '/tmp/test'
phen_list = phenotypes.simulate_traits_w_snps_to_hdf5(
snps, hdf5_file_prefix=file_prefix, num_traits=30, p=0.1)
singletons_thres = []
doubletons_thres = []
common_thres = []
for i, y in enumerate(phen_list['phenotypes']):
K = kinship.calc_ibd_kinship(snps)
K = kinship.scale_k(K)
lmm = lm.LinearMixedModel(y)
lmm.add_random_effect(K)
r1 = lmm.get_REML()
print 'pseudo_heritability:', r1['pseudo_heritability']
ex_res = lm.emmax(snps, y, K)
plt.figure()
plt.hist(y, 50)
plt.savefig('%s_%d_phen.png' % (file_prefix, i))
plt.clf()
agr.plot_simple_qqplots_pvals('%s_%d' % (file_prefix, i), [
ex_res['ps'][:1000], ex_res['ps'][1000:2000], ex_res['ps'][2000:]
],
result_labels=[
'Common SNPs', 'Singletons',
'Doubletons'
],
line_colors=['b', 'r', 'y'],
num_dots=200,
max_neg_log_val=3)
# Cholesky permutations..
res = lm.emmax_perm_test(singleton_snps, y, K, num_perm=1000)
print 1.0 / (20 * 1000.0), res['threshold_05']
singletons_thres.append(res['threshold_05'][0])
res = lm.emmax_perm_test(doubleton_snps, y, K, num_perm=1000)
print 1.0 / (20 * 1000.0), res['threshold_05']
doubletons_thres.append(res['threshold_05'][0])
res = lm.emmax_perm_test(common_snps, y, K, num_perm=1000)
print 1.0 / (20 * 1000.0), res['threshold_05']
common_thres.append(res['threshold_05'][0])
#ATT permutations (Implement)
#PC permutations (Implement)
print sp.mean(singletons_thres), sp.std(singletons_thres)
print sp.mean(doubletons_thres), sp.std(doubletons_thres)
print sp.mean(common_thres), sp.std(common_thres)
