Add codon_prob.py

[willdumm]

Addresses #19

[matsen]

I'm cleaning out old code and came across this, which seems to have been eclipsed but might come in handy at some point:

def codon_change_count(pcp_df):
    # Define a helper function to count nt differences in codons
    def count_nt_differences(codon1, codon2):
        return sum(1 for x, y in zip(codon1, codon2) if x != y)

    # Prepare a list to store results
    results = []

    # Iterate over rows in the DataFrame
    for index, row in pcp_df.iterrows():
        parent_seq = row['parent']
        child_seq = row['child']

        # Ensure sequences are divisible by 3 (length of a codon)
        if len(parent_seq) % 3 != 0 or len(child_seq) % 3 != 0:
            continue  # Skip sequences that are not properly formatted

        # Break sequences into codons
        parent_codons = [parent_seq[i:i+3] for i in range(0, len(parent_seq), 3)]
        child_codons = [child_seq[i:i+3] for i in range(0, len(child_seq), 3)]

        codon_count = len(parent_codons)
        one_nt_change = 0
        two_nt_change = 0
        three_nt_change = 0

        # Count differences for each codon pair
        for pc, cc in zip(parent_codons, child_codons):
            differences = count_nt_differences(pc, cc)
            if differences == 1:
                one_nt_change += 1
            elif differences == 2:
                two_nt_change += 1
            elif differences == 3:
                three_nt_change += 1

        # Append results for this row
        results.append({
            'codon_count': codon_count,
            '1_nt_change': one_nt_change,
            '2_nt_change': two_nt_change,
            '3_nt_change': three_nt_change
        })

    # Convert results list to DataFrame
    return pd.DataFrame(results)

codon_change_df = codon_change_count(pcp_df)

[matsen]

Superseded by #50