SMILES_BindEvaluator/binding_weights.py

import pandas as pd

df = pd.read_csv('/scratch/pranamlab/tong/SMILES_BindEvaluator/datasets/train.csv')

targets = df['Target'].tolist()
total_length = sum([len(seq) for seq in targets])

def parse_sites(x):
    """
    "49,50,51" -> [49,50,51]
    Handles empty/NaN gracefully.
    """
    if x is None:
        return []
    s = str(x).strip()
    if s == "" or s.lower() == "nan":
        return []
    # remove possible quotes
    if (s.startswith('"') and s.endswith('"')) or (s.startswith("'") and s.endswith("'")):
        s = s[1:-1].strip()
    if s == "":
        return []
    return len([int(t.strip()) for t in s.split(",") if t.strip() != ""])

binding_sites = df['Binding Sites'].tolist()
num_binding_sites = sum([parse_sites(site) for site in binding_sites])
num_non_binding_sites = total_length - num_binding_sites

weight_pos =  total_length / (2 * num_binding_sites)
weight_neg = total_length / (2 * num_non_binding_sites)

print(f"Positive Weight: {weight_pos}")
print(f"Negative Weight: {weight_neg}")