samtools calmd -- realignment?

[brentp]

hi, i am trying to understand how the various re-alignment or alignment adjustment tools vary.

from samtools mpileup docs

One way to avoid such an artifact is to do multi-sequence realignment ... SAMtools seeks another solution

but in the samtools calmd help, it lists for option -r:

-r read-independent local realignment

is the disctintion here that tools like SRMA consider the other alignments/reads whereas samtools does a realignment but only considers the reference and the current read?

how does this relate to GATK's Base quality score recalibration?

thanks in advance

[nilshomer]

Samtools does re-alignment (BAQ) via the forward-backward algorithm using the a single read and the reference. SRMA/Dindel/etc use all reads across a given loci. The former really helps identify parts of the alignment that are ambiguous (it could have been an indel or a snp, but I don't know which one), while the latter class of algorithms try to better identify variants from with the read. The BAQ really helps, and so does SRMA/Dindel; it's up to you to choose which you like and in what combination.

GATK's base quality score recalibration is explained in the introduction in the link.

[brentp]

thanks Nils, that makes sense,
actually, i've been using the U87MG decoded paper as a template for our analyses. why didn't (at least by my reading) you guys use SRMA or BAQ. Did the U87MG analysis pre-date those tools?

[nilshomer]

Exactly, the publication pre-dated the tools.