import pandas as pd

# Load the Excel file
df = pd.read_excel("nature19057-SI Table 13.xlsx", sheet_name="Gene Constraint")


# Extract only 'gene' and 'pLI'
df_subset = df[['gene', 'pNull']].copy()

# Rank genes: 1 = highest pLI
df_subset['pNull_rank'] = df_subset['pNull'].rank(ascending=False, method='average')

# Sort by pLI descending (highest first)
df_subset = df_subset.sort_values(by='pNull', ascending=False)

# Write to CSV
df_subset.to_csv("pNull.csv", index=False)

print("Saved to pNull.csv, sorted by descending pNull.")