| 1 | #!/usr/bin/python2.6 |
|---|
| 2 | |
|---|
| 3 | """ |
|---|
| 4 | Mask out potential CpG sites from a maf. Restricted or inclusive definition |
|---|
| 5 | of CpG sites can be used. The total fraction masked is printed to stderr. |
|---|
| 6 | |
|---|
| 7 | usage: %prog < input > output |
|---|
| 8 | -m, --mask=N: Character to use as mask ('?' is default) |
|---|
| 9 | -r, --restricted: Use restricted definition of CpGs |
|---|
| 10 | """ |
|---|
| 11 | |
|---|
| 12 | import bx.align |
|---|
| 13 | import bx.align.maf |
|---|
| 14 | from bx.cookbook import doc_optparse |
|---|
| 15 | import sys |
|---|
| 16 | import bx.align.sitemask.cpg |
|---|
| 17 | |
|---|
| 18 | def main(): |
|---|
| 19 | options, args = doc_optparse.parse( __doc__ ) |
|---|
| 20 | try: |
|---|
| 21 | if options.mask: |
|---|
| 22 | mask = options.mask |
|---|
| 23 | else: |
|---|
| 24 | mask = "?" |
|---|
| 25 | except: |
|---|
| 26 | doc_optparse.exception() |
|---|
| 27 | |
|---|
| 28 | reader = bx.align.maf.Reader( sys.stdin ) |
|---|
| 29 | writer = bx.align.maf.Writer( sys.stdout ) |
|---|
| 30 | |
|---|
| 31 | if options.restricted: |
|---|
| 32 | cpgfilter = bx.align.sitemask.cpg.Restricted( mask=mask ) |
|---|
| 33 | else: |
|---|
| 34 | cpgfilter = bx.align.sitemask.cpg.Inclusive( mask=mask ) |
|---|
| 35 | cpgfilter.run( reader, writer.write ) |
|---|
| 36 | |
|---|
| 37 | print >> sys.stderr, str( float(cpgfilter.masked)/float(cpgfilter.total) * 100 ) + "% bases masked." |
|---|
| 38 | |
|---|
| 39 | if __name__ == "__main__": |
|---|
| 40 | main() |
|---|