|
| 1 | +#!/usr/bin/env python3 |
| 2 | +""" Probabalistically subset FASTA files """ |
| 3 | + |
| 4 | +import argparse |
| 5 | +import os |
| 6 | +import random |
| 7 | +from Bio import SeqIO |
| 8 | + |
| 9 | + |
| 10 | +# -------------------------------------------------- |
| 11 | +def get_args(): |
| 12 | + """get args""" |
| 13 | + parser = argparse.ArgumentParser( |
| 14 | + description='Probabalistically subset FASTA files', |
| 15 | + formatter_class=argparse.ArgumentDefaultsHelpFormatter) |
| 16 | + |
| 17 | + parser.add_argument('file', |
| 18 | + metavar='FILE', |
| 19 | + type=argparse.FileType('r'), |
| 20 | + nargs='+', |
| 21 | + help='Input FASTA file(s)') |
| 22 | + |
| 23 | + parser.add_argument('-p', |
| 24 | + '--pct', |
| 25 | + help='Percent of reads', |
| 26 | + metavar='reads', |
| 27 | + type=float, |
| 28 | + default=.1) |
| 29 | + |
| 30 | + parser.add_argument('-s', |
| 31 | + '--seed', |
| 32 | + help='Random seed value', |
| 33 | + metavar='seed', |
| 34 | + type=int, |
| 35 | + default=None) |
| 36 | + |
| 37 | + parser.add_argument('-o', |
| 38 | + '--outdir', |
| 39 | + help='Output directory', |
| 40 | + metavar='DIR', |
| 41 | + type=str, |
| 42 | + default='out') |
| 43 | + |
| 44 | + args = parser.parse_args() |
| 45 | + |
| 46 | + if not 0 < args.pct < 1: |
| 47 | + parser.error(f'--pct "{args.pct}" must be between 0 and 1') |
| 48 | + |
| 49 | + if not os.path.isdir(args.outdir): |
| 50 | + os.makedirs(args.outdir) |
| 51 | + |
| 52 | + return args |
| 53 | + |
| 54 | + |
| 55 | +# -------------------------------------------------- |
| 56 | +def main(): |
| 57 | + """Make a jazz noise here""" |
| 58 | + |
| 59 | + args = get_args() |
| 60 | + random.seed(args.seed) |
| 61 | + |
| 62 | + total_num = 0 |
| 63 | + for i, fh in enumerate(args.file, start=1): |
| 64 | + basename = os.path.basename(fh.name) |
| 65 | + out_file = os.path.join(args.outdir, basename) |
| 66 | + print(f'{i:3}: {basename}') |
| 67 | + |
| 68 | + out_fh = open(out_file, 'wt') |
| 69 | + num_taken = 0 |
| 70 | + |
| 71 | + for rec in SeqIO.parse(fh, 'fasta'): |
| 72 | + if random.random() <= args.pct: |
| 73 | + num_taken += 1 |
| 74 | + SeqIO.write(rec, out_fh, 'fasta') |
| 75 | + |
| 76 | + out_fh.close() |
| 77 | + total_num += num_taken |
| 78 | + |
| 79 | + num_files = len(args.file) |
| 80 | + print(f'Wrote {total_num:,} sequence{"" if total_num == 1 else "s"} ' |
| 81 | + f'from {num_files:,} file{"" if num_files == 1 else "s"} ' |
| 82 | + f'to directory "{args.outdir}"') |
| 83 | + |
| 84 | + |
| 85 | +# -------------------------------------------------- |
| 86 | +if __name__ == '__main__': |
| 87 | + main() |
0 commit comments