INNER CODE UNIT · Python

preprocess

PharmGKB/PharmCAT · preprocessor/pcat/preprocess.py:47

def preprocess(pharmcat_positions_vcf: Path, reference_genome: Path, regions_to_retain, custom_regions: bool,
               vcf_files: list[Path], samples: list[str], input_basename: str,
               output_dir: Path, output_basename: str = '', split_samples: bool = False,
               keep_intermediate_files: bool = False,
               absent_to_ref: bool = False, unspecified_to_ref: bool = False,
               concurrent_mode: bool = False, max_processes: int = 1, verbose: int = 0) -> list[Path]:
    """
    Normalize and prepare the input VCF for PharmCAT.
    """
    if len(vcf_files) == 0:
        raise ReportableException('Missing VCF input')

    # make sure we have samples
    if samples is None or len(samples) == 0:
        samples = util.read_vcf_samples(vcf_files[0], verbose=verbose)
    else:
        # make sure samples are in the vcf file
        vcf_samples = util.read_vcf_samples(vcf_files[0], verbose=verbose)

View source record →