bed6_to_gtf_TE_annotation.py

#!/usr/bin/env python3

# Author: Jeffrey Grover
# Purpose: Convert a .bed file with 6 columns into a gtf format for TE annotation
# Created: 09/2017

# This script accepts a nonstandard .bed file generated by a repeatmasker
# workflow in the format:
# chr   start   stop    TE-family   TE-type strand
# and outputs a gtf file for use with differential expression workflows

from argparse import ArgumentParser


def bed_to_gtf(input_file, anno_source):
    with open(input_file, 'r') as input_handle:
        for line in input_handle:
            entry = line.split()
            chrom = 'Chr_' + entry[0]
            start = int(entry[1]) + 1  # To convert to 1-based coordinates
            stop = int(entry[2])
            name_family = entry[3]
            te_type = entry[4]
            strand = entry[5]
            print(chrom, anno_source, 'transposable_element', start, stop, '.',
                  strand, '.',
                  'gene_id "%s"; gene_name "%s"; transposable_element_family_ID "%s"; transposable_element_type "%s";'
                  % (name_family, name_family, name_family, te_type), sep='\t')


# Parse command line options

def get_args():
    parser = ArgumentParser(
        description=
        'Converts a .bed file with 6 columns to gtf. It requires that you know the '
        'feature type and source fields and input them using the appropriate '
        'command line option. This was created to convert a bed file for TE '
        'annotations, so use for other purposes will require modification.')
    parser.add_argument('input_path',
                        help='File to process',
                        metavar='FILE.bed')
    parser.add_argument('--source',
                        help='Source of the annotation',
                        type=str)

    return parser.parse_args()


# Process the file

def main(args):
    bed_to_gtf(args.input_path, args.source)


if __name__ == "__main__":
    args = get_args()
    main(args)