bed2gtf

#! /usr/bin/env python3
# -*- coding: utf-8 -*-

__author__="Vivek Rai"
__email__="vivek.rai@roche.com"
__date__="2023-04-28"

# The input should have the following format:
#############################################
# chr1    724805  725050  ...
# chr1    725859  725955  ...

# The output will be named the following way:
# chr1    source     feature     724705  724804   .      +       .      peak_id="chr1_724705_724804"
# chr1    source     feature     724805  725050   .      +       .      peak_id="chr1_724805_725050"
# chr1    source     feature     725051  725150   .      +       .      peak_id="chr1_725051_725150"

import argparse
import sys
import os.path


def run(infile, outfile, args):
    for cont, linea in enumerate(infile):
        linea_split = linea.strip().split()
        chrom = linea_split[0]
        ini_pos = int(linea_split[1])
        fin_pos = int(linea_split[2])

        if len(linea_split) < args.feature_column:
            feature = "peak"
        else:
            feature = linea_split[int(args.feature_column) - 1]

        if not args.strand_column or (len(linea_split) < args.strand_column):
            strand = "."
        else:
            strand = linea_split[int(args.strand_column) - 1]

        print(
            "\t".join(
                [
                    chrom,
                    args.source,
                    feature,
                    str(ini_pos),
                    str(fin_pos),
                    ".",
                    strand,
                    ".",
                    "peak_id=" + chrom + "_" + str(ini_pos) + "_" + str(fin_pos) + ";",
                ]
            ),
            file=outfile,
        )

    outfile.close()
    infile.close()


if __name__ == "__main__":
    parser = argparse.ArgumentParser(description="Converts a bed file to a gtf file.")
    parser.add_argument(
        "-i", "--input", help="File with the regions in bed format", default=sys.stdin
    )
    parser.add_argument(
        "-o", "--output", help="Name of the gtf file output file.", default=sys.stdout
    )
    parser.add_argument(
        "-f",
        "--feature-column",
        help="Column of the bed file with the feature name",
        default=4,
        type=int,
    )
    parser.add_argument("-s", "--source", help="Source of the data", default="bed2gtf")
    parser.add_argument(
        "-t",
        "--strand-column",
        help="Column of the bed file with strand info",
        default=None,
        type=int,
    )

    args = parser.parse_args()

    if args.input == sys.stdin:
        infile = sys.stdin
    else:
        if not os.path.isfile(args.input):
            print("Error: Input file does not exist")
            sys.exit(1)
        infile = open(args.input, "r")

    if args.output == sys.stdout:
        outfile = sys.stdout
    else:
        outfile = open(args.output, "w")

    run(infile, outfile, args)