Multiple alignments for a read not handled properly

[GoogleCodeExporter]

Using -r 1 (the default), and a read has multiple placements, bsmap selects one 
of the placements at random and marks it as a non-primary alignment.  Marking 
as non-primary isn't the right thing to do.  Rather, the read should get 
MAPQ=0.  Non-primary flag is only appropriate when multiple alignments have 
been written to the output.  The convention of MAPQ=0 implying that there were 
multiple alignments but one was selected randomly appears to have been dropped 
from the spec, but it is the standard.

The attached patch makes this change.

Thanks, Alec

Original issue reported on code.google.com by al...@broadinstitute.org on 18 Dec 2012 at 2:46

Attachments:

• bsmap.patch