dbretina / dbretina Goto Github PK

def vcf_to_assoc(vcf_filename, assoc_filename):

    # Open the VCF file using a suitable library (e.g., pysam)
    vcf_file = open_vcf_file(vcf_filename)

    # Open the output file in write mode
    with open(assoc_filename, 'w') as assoc_file:

        # Loop over each record in the VCF file
        for record in vcf_file:

            # For each sample in this record
            for sample in record.samples:

                # Check if this sample has a variant at this position (i.e., genotype is not (0, 0))
                if sample_has_variant(sample):

                    # Prepare a line with sample ID and variant details
                    line = format_line(sample, record)

                    # Write this line to the association file
                    assoc_file.write(line)