from Bio import SeqIO
from collections import Counter

def calculate_complementary_ratios(sequence):
    complementary_base = {'A': 'T', 'T': 'A', 'C': 'G', 'G': 'C'}
    counts = Counter(sequence)
    ratios = {base: counts[base] / len(sequence) for base in 'ATCG'}
    complementary_ratios = {base: ratios[complementary_base[base]] for base in 'ATCG'}
    return complementary_ratios

def fastq_summary(input_fastq, output_summary):
    with open(input_fastq, "r") as handle, open(output_summary, "w") as output:
        for record in SeqIO.parse(handle, "fastq"):
            seq_id = record.id
            sequence = str(record.seq)
            quality_scores = record.letter_annotations["phred_quality"]

            # Calculate metrics
            counts = Counter(sequence)
            frequencies = {base: counts[base] for base in 'ATCG'}
            complementary_ratios = calculate_complementary_ratios(sequence)
            mean_quality_score = sum(quality_scores) / len(quality_scores)

            # Write metrics to the output file
            output.write(f"Sequence ID: {seq_id}\n")
            output.write(f"Total Counts: {counts}\n")
            output.write(f"Frequencies: {frequencies}\n")
            output.write(f"Complementary Ratios: {complementary_ratios}\n")
            output.write(f"Mean Quality Score: {mean_quality_score:.2f}\n\n")

if __name__ == "__main__":
    input_fastq = "8_Swamp_S1B_MATK_2019_minq7.fastq"
    output_summary = "fastq_summary.txt"
    fastq_summary(input_fastq, output_summary)