def continuous_enrichment(pheno,
data,
covariate,
node_labels='node_name_not_specificed'):
sub_mask = find_subset(pheno, covariate)
sub_pheno = pheno.loc[sub_mask]
sub_data = data[sub_mask, :]
n_data = sub_data.shape[1]
results = {key: list() for key in ['node', 'pearson_r', 'p_value']}
for node_id in range(n_data):
r, p = stats.pearsonr(sub_pheno[covariate], sub_data[:, node_id])
# Store the results
results['node'].append(node_id + 1)
results['pearson_r'].append(r)
results['p_value'].append(p)
results_table = pd.DataFrame(data=results)
# Correct for multiple comparisons
(fdr_pass, qval, _, _) = stm(results_table.p_value,
alpha=0.05,
method='fdr_bh')
results_table['q_value'] = qval
# Add the node_names
results_table['node_name'] = node_labels
return results_table
def summarize_glm(glm_table, conn_mask, roi_labels):
out_table = glm_table.copy()
(fdr_pass, qval, _, _) = stm(glm_table.pvals, alpha=0.05, method='fdr_bh')
out_table['qval'] = qval
# Return to matrix form
stand_beta_table = pd.DataFrame(conn2mat(out_table.stand_betas.values,
conn_mask),
index=roi_labels,
columns=roi_labels)
qval_table = pd.DataFrame(conn2mat(out_table.pvals.values, conn_mask),
index=roi_labels,
columns=roi_labels)
return out_table, stand_beta_table, qval_table
# Determine if cases > controls or reverse
case_gt_con = u_right > u_min
if not case_gt_con:
r_b = -r_b
# Store the results
results['node'].append(node_id + 1)
results['U'].append(u_min)
results['p_value'].append(p)
results[f'median_{group}_{case}'].append(median_case)
results[f'median_{group}_{control}'].append(median_control)
results['rank_biserial_correlation'].append(r_b)
results_table = pd.DataFrame(data=results)
# Correct for multiple comparisons
(fdr_pass, qval, _, _) = stm(results_table.p_value,
alpha=0.05,
method='fdr_bh')
results_table['q_value'] = qval
# Add the node_names
results_table['node_name'] = node_labels
return results_table
def continuous_enrichment(pheno,
data,
covariate,
node_labels='node_name_not_specificed'):
sub_mask = find_subset(pheno, covariate)
sub_pheno = pheno.loc[sub_mask]
sub_data = data[sub_mask, :]