Not Using Marked Duplicates In Somatic Sniper

Question

[jaime.davilal]

Hi.

I was wondering if SomaticSniper considers marked duplicate reads in calling somatic variants and from the options and code it is my impression that it does.

Would a strategy similar to the one described in:

http://seqanswers.com/forums/showthread.php?t=22752, work in order not to consider the marked duplicated reads?

Namely in sniper_pileup.c in the function

   bam_plbuf_t *bam_plbuf_init(bam_pileup_f func, void *data)

to include the line

    buf->flag_mask &= BAM_FDUP;

before

     return buf;

Would this be a good feature to add in somatic sniper?

Thanks

Jaime

[Istvan Albert]

to be honest this is something that only the developer can really comment on - the rest of us here are happy if a tool works in the first place - let alone mess with its innards :-)

[Istvan Albert]

also you may want to use the feature called request help via email above to send this question to the developer

Answer 1

[ernfrid]

Hi Jaime, This is a good question. As one of the developers, I can tell you that SomaticSniper does not consider marked duplicates, secondary mappings, qc failed reads, or unmapped reads. The reason it does not is that it uses the exact same flag that samtools mpileup uses as its default. For reference, that flag is listed below (from the samtools source code):

/*! @abstract defautl mask for pileup */
#define BAM_DEF_MASK (BAM_FUNMAP | BAM_FSECONDARY | BAM_FQCFAIL | BAM_FDUP)

This flag is set in the main sniper function (main.c) as an argument to bamsspileupfile (d->mask).

Dave