[2] | 1 | #!/usr/bin/env python |
---|
| 2 | |
---|
| 3 | """ |
---|
| 4 | Convert from interval file to interval index file. Default input file format is BED (0-based, half-open intervals). |
---|
| 5 | |
---|
| 6 | usage: %prog in_file out_file |
---|
| 7 | -G, --gff: input is GFF format, meaning start and end coordinates are 1-based, closed interval |
---|
| 8 | """ |
---|
| 9 | |
---|
| 10 | from __future__ import division |
---|
| 11 | |
---|
| 12 | import sys, fileinput |
---|
| 13 | from galaxy import eggs |
---|
| 14 | import pkg_resources; pkg_resources.require( "bx-python" ) |
---|
| 15 | from galaxy.visualization.tracks.summary import * |
---|
| 16 | from bx.cookbook import doc_optparse |
---|
| 17 | from galaxy.tools.util.gff_util import convert_gff_coords_to_bed |
---|
| 18 | from bx.interval_index_file import Indexes |
---|
| 19 | |
---|
| 20 | def main(): |
---|
| 21 | |
---|
| 22 | # Read options, args. |
---|
| 23 | options, args = doc_optparse.parse( __doc__ ) |
---|
| 24 | try: |
---|
| 25 | gff_format = bool( options.gff ) |
---|
| 26 | input_fname, out_fname = args |
---|
| 27 | except: |
---|
| 28 | doc_optparse.exception() |
---|
| 29 | |
---|
| 30 | # Do conversion. |
---|
| 31 | # TODO: take column numbers from command line. |
---|
| 32 | if gff_format: |
---|
| 33 | chr_col, start_col, end_col = ( 0, 3, 4 ) |
---|
| 34 | else: |
---|
| 35 | chr_col, start_col, end_col = ( 0, 1, 2 ) |
---|
| 36 | index = Indexes() |
---|
| 37 | offset = 0 |
---|
| 38 | for line in open(input_fname, "r"): |
---|
| 39 | feature = line.strip().split() |
---|
| 40 | if not feature or feature[0].startswith("track") or feature[0].startswith("#"): |
---|
| 41 | offset += len(line) |
---|
| 42 | continue |
---|
| 43 | chrom = feature[ chr_col ] |
---|
| 44 | chrom_start = int( feature[ start_col ] ) |
---|
| 45 | chrom_end = int( feature[ end_col ] ) |
---|
| 46 | if gff_format: |
---|
| 47 | chrom_start, chrom_end = convert_gff_coords_to_bed( [chrom_start, chrom_end ] ) |
---|
| 48 | index.add( chrom, chrom_start, chrom_end, offset ) |
---|
| 49 | offset += len(line) |
---|
| 50 | |
---|
| 51 | index.write( open(out_fname, "w") ) |
---|
| 52 | |
---|
| 53 | if __name__ == "__main__": |
---|
| 54 | main() |
---|
| 55 | |
---|