gen-collection-for-merged-fasta.py

#!/usr/bin/env python
# To use this:
# python gen-collection-for-merged-fasta.py -f fasta.txt -o collection-file.txt

import sys

from anvio import fastalib
from anvio import utils
from anvio import filesnpaths
from anvio.errors import ConfigError, FilesNPathsError


def main(args):
    fasta_files = utils.get_TAB_delimited_file_as_dictionary(args.fasta_txt)

    with open(args.output_file, 'w') as f:
        for name in fasta_files:
            fasta_reader = fastalib.ReadFasta(fasta_files[name]['path'])
            contigs = fasta_reader.ids
            for c in contigs:
                f.write('%s\t%s\n' % (c, name))


if __name__ == '__main__':
    import argparse
    parser = argparse.ArgumentParser(description="Takes a fasta.txt file, and returns a collection file for contigs.")

    parser.add_argument('-f', '--fasta-txt', metavar = 'FILE', help = 'fasta.txt file (the same as used in anvio contigs snakemake workflow)')
    parser.add_argument('-o', '--output-file', metavar = 'FILE', help = 'Output file name.')

    args = parser.parse_args()

    try:
        main(args)
    except ConfigError as e:
        print(e)
        sys.exit(-1)
    except FilesNPathsError as e:
        print(e)
        sys.exit(-2)