def all_tf_corrected_pvalues():
import seaborn as sns
sns.set_style("dark")
from limix.plot import qqplot
print("Loading pvalues")
all_tf_corrected_results = pickle.load(
open(
GTEx_directory +
'/results/InflationPvalues/all_tf_corrected_pvalues.pickle',
'rb'))
tf_corrected_lamb = estimate_lambda(
all_tf_corrected_results[0].flatten())
qqplot(all_tf_corrected_results[0].flatten())
qqplot(all_tf_corrected_results[1].flatten())
plt.title('$\lambda={:0.2f}'.format(tf_corrected_lamb))
os.makedirs(GTEx_directory + '/plotting/{}'.format(group),
exist_ok=True)
plt.savefig(
GTEx_directory +
'/plotting/{group}/{name}.eps'.format(group=group, name=name),
format='eps',
dpi=100)
plt.savefig(
GTEx_directory +
'/plotting/{group}/{name}.png'.format(group=group, name=name),
format='png',
dpi=100)
plt.show()
def pc_corrected_pvalues():
print('Loading corrected pvalues')
pc_corrected_results = pickle.load(
open(
GTEx_directory +
'/results/InflationPvalues/pc_corrected_pvalues.pickle', 'rb'))
import seaborn as sns
sns.set_style("dark")
from limix.plot import qqplot
for (i, pvalues) in enumerate(pc_corrected_results):
Rs_real, pvs_real, pvs_1, pvs_2, pvs_3 = pvalues
print('Calculating lambda for PC {}'.format(i))
lamb = estimate_lambda(pvs_real.flatten())
print('Plotting PC {}'.format(i))
qqplot(pvs_real.flatten(),
label=r'{} PCs, $\lambda={:0.2f}$'.format(i + 1, lamb))
plt.legend(prop={'size': 15}, loc='upper left')
os.makedirs(GTEx_directory + '/plotting/{}'.format(group),
exist_ok=True)
plt.savefig(
GTEx_directory +
'/plotting/{group}/{name}.eps'.format(group=group, name=name),
format='eps',
dpi=100)
plt.savefig(
GTEx_directory +
'/plotting/{group}/{name}.png'.format(group=group, name=name),
format='png',
dpi=100)
plt.show()
def raw_pvalues():
import seaborn as sns
sns.set_style("dark")
results = pickle.load(
open(
GTEx_directory + '/results/{group}/{name}.pickle'.format(
group=group, name=name), 'rb'))
Rs_real, pvs_real, pvs_1, pvs_2, pvs_3 = results
from limix.plot import qqplot
qqplot(pvs_real.flatten())
qqplot(pvs_1.flatten())
qqplot(pvs_2.flatten())
qqplot(pvs_3.flatten())
os.makedirs(GTEx_directory + '/plotting/{}'.format(group),
exist_ok=True)
plt.savefig(
GTEx_directory +
'/plotting/{group}/{name}.eps'.format(group=group, name=name),
format='eps',
dpi=100)
plt.savefig(
GTEx_directory +
'/plotting/{group}/{name}.png'.format(group=group, name=name),
format='png',
dpi=100)
plt.show()
def tf_vs_pc_corrected_pvalues():
print("Loading pvalues")
pc_corrected_results = pickle.load(
open(
GTEx_directory +
'/results/InflationPvalues/pc_corrected_pvalues.pickle', 'rb'))
tf_corrected_results = pickle.load(
open(
GTEx_directory +
'/results/InflationPvalues/tf_corrected_pvalues.pickle', 'rb'))
import seaborn as sns
sns.set_style("dark")
from limix.plot import qqplot
_, pvs_real_pc_corrected, _, _, _ = pc_corrected_results[0]
_, pvs_real_tf_corrected, _, _, _ = tf_corrected_results
print('Estimating lambda for raw pvalues')
pc_corrected_lamb = estimate_lambda(pvs_real_pc_corrected.flatten())
print('Estimating lambda for TF corrected pvalues')
tf_corrected_lamb = estimate_lambda(pvs_real_tf_corrected.flatten())
print('Plotting 1 PC corrected pvalues')
qqplot(
pvs_real_pc_corrected.flatten(),
label='1 PC Corrected $\lambda={:0.2f}$'.format(pc_corrected_lamb))
print('Plotting TF corrected pvalues')
qqplot(
pvs_real_tf_corrected.flatten(),
label='5 TF Corrected $\lambda={:0.2f}$'.format(tf_corrected_lamb))
plt.legend(prop={'size': 15})
os.makedirs(GTEx_directory + '/plotting/{}'.format(group),
exist_ok=True)
plt.savefig(
GTEx_directory +
'/plotting/{group}/{name}.eps'.format(group=group, name=name),
format='eps',
dpi=100)
plt.savefig(
GTEx_directory +
'/plotting/{group}/{name}.png'.format(group=group, name=name),
format='png',
dpi=100)
plt.show()
for pv in ['pv10', 'pv20', 'pv21']:
test = tests.test[pv]
# plot results
# Manhattan plot
gwas_pv = gwas_results[['chrom', 'pos', pv]]
gwas_pv.columns = ['chrom', 'pos', 'pv']
plot.manhattan(gwas_pv)
plt = plot.get_pyplot()
plt.savefig(
f"{args.outDir}/manhattanPlot_{'_'.join(traits)}_{args.maf}_{test}.png"
)
plt.close()
# QQ-plot
plot.qqplot(gwas_pv.pv)
plt = plot.get_pyplot()
plt.savefig(
f"{args.outDir}/qqPlot_{'_'.join(traits)}_{args.maf}_{test}.png")
plt.close()
# save results
gwas_pv['maf'] = SNPs_MAF
gwas_pv['mac'] = gwas_pv.maf * len(acn_indices)
gwas_pv.mac = gwas_pv.mac.astype(int)
gwas_pv['GVE'] = np.nan
gwas_pv.columns.values[0] = 'chr'
gwas_pv.columns.values[2] = 'pvalue'
gwas_pv.to_csv(
f"{args.outDir}/{'_'.join(traits)}_{args.maf}_MTMM_{test}.csv",
index=False)
gwas_results["pos"] = gwas_results['pos'].astype('int')
gwas_results["pv"] = gwas_results['pv'].astype('float')
gwas_results["GVE"] = gwas_results['GVE'].astype('float')
#gwas_results.dtypes
Bonferroni = multitest.multipletests(pvalues, alpha = 0.05, method = 'fdr_bh')[3]
plot.manhattan(gwas_results)
plt = plot.get_pyplot()
_ = plt.axhline(-math.log(Bonferroni, 10), color='red')
plt.savefig(f'{output_file}.manhattan.png')
plt.close()
###########################################
### 10. QQplot
plot.qqplot(gwas_results.pv)
plt = plot.get_pyplot()
plt.savefig(f'{output_file}.QQplot.png')
plt.close()
###########################################
### 11. Calculate heritability
from numpy.random import RandomState
from limix.her import estimate
herit = estimate(y=Y, K=K, lik="normal", verbose=False)
print(herit)
f = open(f"{output_file}.herit.txt", "w")
f.write(str(herit))
x.query("fdr > 0.05").sort_values("r2", ascending=False).head(60)
#
#
for dtype, lm_df in [("Drug-Protein", lm_drug), ("CRISPR-Protein", lm_crispr)]:
for pvar in ["nc_pval", "pval"]:
# dtype, lm_df, pvar = "Drug-Protein", lm_drug, "nc_pval"
# _, ax = plt.subplots(1, 1, figsize=(3, 3), dpi=600)
plot_df = []
for n in PPI_ORDER:
df = lm_df.query(f"ppi == '{n}'")
x_var, y_var, l_var = qqplot(
df[pvar],
label=n,
# ax=ax,
show_lambda=True,
band_kws=dict(color=PPI_PAL[n], alpha=0.2),
pts_kws=dict(color=PPI_PAL[n]),
)
plot_df.append(
pd.DataFrame({
"qnull": x_var,
"qemp": y_var,
"lambda": l_var,
"target": n
}))
plt.close("all")
plot_df = pd.concat(plot_df)
