# Filter the dimer confidence score list output by dimer_confidence.py
# to list just those with confidence above a specified percentage of
# interface residue pairs with good PAE values, listing only highest
# scoring of 5 alphafold models.

def high_confidence_dimers(lines, cutoff = 0.10):
    scores = {}
    best = {}
    for line in lines:
        fields = line.split(',')
        pconf = float(fields[7])
        name_end = line.find('_unrelaxed')
        genes = line[:name_end]
        if pconf >= cutoff and (genes not in scores or scores[genes] < pconf):
            scores[genes] = pconf
            best[genes] = line.strip()
    best_dimers = list(best.values())
    best_dimers.sort()
    return best_dimers

with open('dimer_confidence.csv', 'r') as file:
    lines = file.readlines()

print('\n'.join(high_confidence_dimers(lines, cutoff = 0.1)))