Gradient correlation

Find genes that could be PCGs by calculating and evaluating correlation coefficient values between known PCGs (from published studies) and target genes.

Calculating Spearman’s rank correlation coefficient

import scipy.stats
import scanpy as sc
import pandas as pd

adata = sc.read_h5ad(file_name)
df = adata.to_df()
target_genes = list(df.columns)  # or assigned by users
with open('known_pcgs.txt', 'r') as f:
  known_pcgs = f.read().splitlines()

Finding out more polarity-related genes

df_data = {'gene':[], 'pcg':[], 'scc':[]}
for gene in target_genes:
  for pcg in known_pcgs:
    if pcg in df.columns and pcg != gene:
      df_data['gene'].append(gene)
      df_data['pcg'].append(pcg)
      scc = scipy.stats.spearmanr(df[gene], df[pcg]).correlation
      df_data['scc'].append(scc)

results = pd.DataFrame(df_data)
results = results.sort_values(by='scc', ascending=False)
results.to_csv('spearman.csv', index=False)