fastq_to_fastq.py 文件源码

def get_arguments():
    parser = argparse.ArgumentParser(description='FASTQ filter tool',
                                     formatter_class=argparse.ArgumentDefaultsHelpFormatter)
    parser.add_argument('input_fastq', type=str,
                        help='FASTQ file of reads to be filtered (can be gzipped)')
    parser.add_argument('--min_length', type=int, default=0,
                        help='Exclude reads shorter than this length (in bp)')
    parser.add_argument('--min_mean_qual', type=float, default=0.0,
                        help='Exclude reads with a mean qscore less than this value')
    parser.add_argument('--min_qual_window', type=float, default=0.0,
                        help='Exclude reads where their mean qscore in a sliding window drops '
                             'below this value')
    parser.add_argument('--window_size', type=int, default=50,
                        help='The size of the sliding window used for --min_qual_window')
    parser.add_argument('--target_bases', type=int, default=None,
                        help='If set, exclude the worst reads (as judged by their minimum qscore '
                             'in a sliding window) such that only this many bases remain')
    args = parser.parse_args()

    args.input_fastq = os.path.abspath(args.input_fastq)
    if not os.path.isfile(args.input_fastq):
        sys.exit('Error: could not find ' + args.input_fastq)

    if args.min_length == 0 and args.min_mean_qual == 0.0 and args.min_qual_window == 0.0 and \
            args.target_bases is None:
        sys.exit('Error: no filters were used so this tool refuses to run (because the output\n'
                 '       FASTQ would be identical to the input FASTQ). Please use one of the\n'
                 '       following filters: --min_length, --min_mean_qual, --min_qual_window\n'
                 '       or --target_bases.')
    return args