| 1 | #!/usr/bin/python2.6 |
|---|
| 2 | |
|---|
| 3 | """ |
|---|
| 4 | Convert wiggle data to a binned array. This assumes the input data is on a |
|---|
| 5 | single chromosome and does no sanity checks! |
|---|
| 6 | |
|---|
| 7 | usage: %prog score_file out_file < wiggle_data |
|---|
| 8 | -c, --comp=type: compression type (none, zlib, lzo) |
|---|
| 9 | """ |
|---|
| 10 | |
|---|
| 11 | from __future__ import division |
|---|
| 12 | |
|---|
| 13 | import sys |
|---|
| 14 | import psyco_full |
|---|
| 15 | import bx.wiggle |
|---|
| 16 | from bx.binned_array import BinnedArray |
|---|
| 17 | from bx_extras.fpconst import isNaN |
|---|
| 18 | from bx.cookbook import doc_optparse |
|---|
| 19 | from bx import misc |
|---|
| 20 | |
|---|
| 21 | def main(): |
|---|
| 22 | |
|---|
| 23 | # Parse command line |
|---|
| 24 | options, args = doc_optparse.parse( __doc__ ) |
|---|
| 25 | try: |
|---|
| 26 | if options.comp: |
|---|
| 27 | comp_type = options.comp |
|---|
| 28 | else: |
|---|
| 29 | comp_type = None |
|---|
| 30 | score_fname = args[0] |
|---|
| 31 | out_fname = args[1] |
|---|
| 32 | except: |
|---|
| 33 | doc_optparse.exit() |
|---|
| 34 | |
|---|
| 35 | scores = BinnedArray() |
|---|
| 36 | |
|---|
| 37 | ## last_chrom = None |
|---|
| 38 | for i, ( chrom, pos, val ) in enumerate( bx.wiggle.Reader( misc.open_compressed( score_fname ) ) ): |
|---|
| 39 | #if last_chrom is None: |
|---|
| 40 | # last_chrom = chrom |
|---|
| 41 | #else: |
|---|
| 42 | # assert chrom == last_chrom, "This script expects a 'wiggle' input on only one chromosome" |
|---|
| 43 | scores[pos] = val |
|---|
| 44 | # Status |
|---|
| 45 | if i % 10000 == 0: print i, "scores processed" |
|---|
| 46 | |
|---|
| 47 | out = open( out_fname, "w" ) |
|---|
| 48 | if comp_type: |
|---|
| 49 | scores.to_file( out, comp_type=comp_type ) |
|---|
| 50 | else: |
|---|
| 51 | scores.to_file( out ) |
|---|
| 52 | out.close() |
|---|
| 53 | |
|---|
| 54 | if __name__ == "__main__": main() |
|---|