[2] | 1 | #!/usr/bin/env python |
---|
| 2 | |
---|
| 3 | """ |
---|
| 4 | Condenses pileup format into ranges of bases. |
---|
| 5 | |
---|
| 6 | usage: %prog [options] |
---|
| 7 | -i, --input=i: Input pileup file |
---|
| 8 | -o, --output=o: Output pileup |
---|
| 9 | -c, --coverage=c: Coverage |
---|
| 10 | -f, --format=f: Pileup format |
---|
| 11 | -b, --base=b: Base to select |
---|
| 12 | -s, --seq_column=s: Sequence column |
---|
| 13 | -l, --loc_column=l: Base location column |
---|
| 14 | -r, --base_column=r: Reference base column |
---|
| 15 | -C, --cvrg_column=C: Coverage column |
---|
| 16 | """ |
---|
| 17 | |
---|
| 18 | from galaxy import eggs |
---|
| 19 | import pkg_resources; pkg_resources.require( "bx-python" ) |
---|
| 20 | from bx.cookbook import doc_optparse |
---|
| 21 | import sys |
---|
| 22 | |
---|
| 23 | def stop_err( msg ): |
---|
| 24 | sys.stderr.write( msg ) |
---|
| 25 | sys.exit() |
---|
| 26 | |
---|
| 27 | def __main__(): |
---|
| 28 | strout = '' |
---|
| 29 | #Parse Command Line |
---|
| 30 | options, args = doc_optparse.parse( __doc__ ) |
---|
| 31 | coverage = int(options.coverage) |
---|
| 32 | fin = file(options.input, 'r') |
---|
| 33 | fout = file(options.output, 'w') |
---|
| 34 | inLine = fin.readline() |
---|
| 35 | if options.format == 'six': |
---|
| 36 | seqIndex = 0 |
---|
| 37 | locIndex = 1 |
---|
| 38 | baseIndex = 2 |
---|
| 39 | covIndex = 3 |
---|
| 40 | elif options.format == 'ten': |
---|
| 41 | seqIndex = 0 |
---|
| 42 | locIndex = 1 |
---|
| 43 | if options.base == 'first': |
---|
| 44 | baseIndex = 2 |
---|
| 45 | else: |
---|
| 46 | baseIndex = 3 |
---|
| 47 | covIndex = 7 |
---|
| 48 | else: |
---|
| 49 | seqIndex = int(options.seq_column) - 1 |
---|
| 50 | locIndex = int(options.loc_column) - 1 |
---|
| 51 | baseIndex = int(options.base_column) - 1 |
---|
| 52 | covIndex = int(options.cvrg_column) - 1 |
---|
| 53 | lastSeq = '' |
---|
| 54 | lastLoc = -1 |
---|
| 55 | locs = [] |
---|
| 56 | startLoc = -1 |
---|
| 57 | bases = [] |
---|
| 58 | while inLine.strip() != '': |
---|
| 59 | lineParts = inLine.split('\t') |
---|
| 60 | try: |
---|
| 61 | seq, loc, base, cov = lineParts[seqIndex], int(lineParts[locIndex]), lineParts[baseIndex], int(lineParts[covIndex]) |
---|
| 62 | except IndexError, ei: |
---|
| 63 | if options.format == 'ten': |
---|
| 64 | stop_err( 'It appears that you have selected 10 columns while your file has 6. Make sure that the number of columns you specify matches the number in your file.\n' + str( ei ) ) |
---|
| 65 | else: |
---|
| 66 | stop_err( 'There appears to be something wrong with your column index values.\n' + str( ei ) ) |
---|
| 67 | except ValueError, ev: |
---|
| 68 | if options.format == 'six': |
---|
| 69 | stop_err( 'It appears that you have selected 6 columns while your file has 10. Make sure that the number of columns you specify matches the number in your file.\n' + str( ev ) ) |
---|
| 70 | else: |
---|
| 71 | stop_err( 'There appears to be something wrong with your column index values.\n' + str( ev ) ) |
---|
| 72 | # strout += str(startLoc) + '\n' |
---|
| 73 | # strout += str(bases) + '\n' |
---|
| 74 | # strout += '%s\t%s\t%s\t%s\n' % (seq, loc, base, cov) |
---|
| 75 | if loc == lastLoc+1 or lastLoc == -1: |
---|
| 76 | if cov >= coverage: |
---|
| 77 | if seq == lastSeq or lastSeq == '': |
---|
| 78 | if startLoc == -1: |
---|
| 79 | startLoc = loc |
---|
| 80 | locs.append(loc) |
---|
| 81 | bases.append(base) |
---|
| 82 | else: |
---|
| 83 | if len(bases) > 0: |
---|
| 84 | fout.write('%s\t%s\t%s\t%s\n' % (lastSeq, startLoc-1, lastLoc, ''.join(bases))) |
---|
| 85 | startLoc = loc |
---|
| 86 | locs = [loc] |
---|
| 87 | bases = [base] |
---|
| 88 | else: |
---|
| 89 | if len(bases) > 0: |
---|
| 90 | fout.write('%s\t%s\t%s\t%s\n' % (lastSeq, startLoc-1, lastLoc, ''.join(bases))) |
---|
| 91 | startLoc = -1 |
---|
| 92 | locs = [] |
---|
| 93 | bases = [] |
---|
| 94 | else: |
---|
| 95 | if len(bases) > 0: |
---|
| 96 | fout.write('%s\t%s\t%s\t%s\n' % (lastSeq, startLoc-1, lastLoc, ''.join(bases))) |
---|
| 97 | if cov >= coverage: |
---|
| 98 | startLoc = loc |
---|
| 99 | locs = [loc] |
---|
| 100 | bases = [base] |
---|
| 101 | else: |
---|
| 102 | startLoc = -1 |
---|
| 103 | locs = [] |
---|
| 104 | bases = [] |
---|
| 105 | lastSeq = seq |
---|
| 106 | lastLoc = loc |
---|
| 107 | inLine = fin.readline() |
---|
| 108 | if len(bases) > 0: |
---|
| 109 | fout.write('%s\t%s\t%s\t%s\n' % (lastSeq, startLoc-1, lastLoc, ''.join(bases))) |
---|
| 110 | fout.close() |
---|
| 111 | fin.close() |
---|
| 112 | |
---|
| 113 | # import sys |
---|
| 114 | # strout += file(fout.name,'r').read() |
---|
| 115 | # sys.stderr.write(strout) |
---|
| 116 | |
---|
| 117 | if __name__ == "__main__" : __main__() |
---|