1 | #!/usr/bin/env python |
---|
2 | |
---|
3 | """ |
---|
4 | Convert from interval file to summary tree 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 | from __future__ import division |
---|
10 | |
---|
11 | import sys, fileinput |
---|
12 | from galaxy import eggs |
---|
13 | import pkg_resources; pkg_resources.require( "bx-python" ) |
---|
14 | from galaxy.visualization.tracks.summary import * |
---|
15 | from bx.intervals.io import * |
---|
16 | from bx.cookbook import doc_optparse |
---|
17 | from galaxy.tools.util.gff_util import GFFReaderWrapper |
---|
18 | |
---|
19 | def main(): |
---|
20 | # Read options, args. |
---|
21 | options, args = doc_optparse.parse( __doc__ ) |
---|
22 | try: |
---|
23 | gff_format = bool( options.gff ) |
---|
24 | input_fname, out_fname = args |
---|
25 | except: |
---|
26 | doc_optparse.exception() |
---|
27 | |
---|
28 | # Do conversion. |
---|
29 | # TODO: take column numbers from command line. |
---|
30 | if gff_format: |
---|
31 | reader_wrapper_class = GFFReaderWrapper |
---|
32 | chr_col, start_col, end_col, strand_col = ( 0, 3, 4, 6 ) |
---|
33 | else: |
---|
34 | reader_wrapper_class = NiceReaderWrapper |
---|
35 | chr_col, start_col, end_col, strand_col = ( 0, 1, 2, 5 ) |
---|
36 | reader_wrapper = reader_wrapper_class( fileinput.FileInput( input_fname ), |
---|
37 | chrom_col=chr_col, |
---|
38 | start_col=start_col, |
---|
39 | end_col=end_col, |
---|
40 | strand_col=strand_col, |
---|
41 | fix_strand=True ) |
---|
42 | st = SummaryTree(block_size=25, levels=6, draw_cutoff=150, detail_cutoff=30) |
---|
43 | for line in list( reader_wrapper ): |
---|
44 | if type( line ) is GenomicInterval: |
---|
45 | st.insert_range( line.chrom, long( line.start ), long( line.end ) ) |
---|
46 | |
---|
47 | st.write(out_fname) |
---|
48 | |
---|
49 | if __name__ == "__main__": |
---|
50 | main() |
---|