non-unique matches

[Perry]

Hello-

Just to be clear, the -T option (Maximum number of matches) controls
reads which align with equal scores to multiple locations in the
reference sequence, correct? So if I only want reads which uniquely
align to a single location in the reference sequence I can set this to
1, correct?

Also, for those reads which do not uniquely align, are they flagged in
the sam file? For example, BWA has a custom tag, X0:i:n, where n is
the number of different places the read aligns.

Thanks!

[Nathan Clement]

The -T option only limits the number of *similar* matches, not the number of locations with equal scores. With the -a option, you specify a minimum alignment score. If a read matches too many locations within this threshold, it will be considered noise and not used.

If this is helpful, we can add it to GNUMAP. If it is, are you only concerned with the SAM file, or should this be incorporated into the rest of the mapping process, such as SNP calling and the .sgr file?

Nathan

[Perry]

Okay, -T understood.

Knowing which reads do not uniquely align would be very help, actually
essential, if we are to use GNUMAP. The reason is that in addition to
our research, we are going to be running clinical samples for
diagnostics. I realize that this will make it so that some regions
likely have low coverage; however, from a clinical standpoint it's
better to Sanger sequence regions of low coverage than potentially
have coverage from misaligned reads. So it would be very helpful for
me, but NGS is not widely used in clinical settings and so I'm not
sure it's worth the effort for a single user. The preference would
definitely be seeing this information in both alignment and variant
files.

Also wondering...have you considered outputting variants in VCF or GVF
formats? VCF is very common, it's the format used by the 1000 Genomes
project. GVF is a new format which is more intuitive to use and is
being used instead of the VCF format in at least a couple applications
I'm aware of. Just curious.