How Do You Do Local Realignment Around Indels For Exome Sequencing?

Question

[Travis]

Hi all,

I am wondering how people generally go about local realignment around indels in their workflow?

The most popular programs seem to be SRMA and GATK - does anyone use anything else?

In my initial tests this seems to be a VERY time-consuming step when compared to the initial alignment step. Does anyone employ methods of speeding up the process? For example, only realigning within exonic coordinates OR perhaps performing variant calling first and then using the coordinates from the VCF to restrict the areas where realignment is performed?

I am trying to set up a SNP/Indel workflow and realignment seems to slow it to the the point where it becomes useless for anything but small numbers of samples. I tested on a single chromosome and it took 4.5 hours. Alignment and variant calling without realignment only took about 1.5 hours for the whole genome!

Thanks in advance!

[Biomonika (Noolean)]

Did you find anything new in past months? I am now facing similiar situation as you were, considering which tool (if any) to use. Thanks.

Answer 1

[Drio]

Take a look to this question and my answer (and the comments). SRMA allows realignment of specific regions. That will help you reduce the runtime of the process.

[Travis]

I remember the question - I asked it :) My question here is a more general one.

Answer 2

[Docroberson]

I typically use BWA for alignments to get indels at the same time. Pretty fast (on par with bowtie). Alternatively I've used NOVOALIGN to do alignment and indel calculations in one step also. Very good at finding both substitutions and indels, but VERY slow and resource intensive compared to other software.

[Travis]

I am more curious about the local realignment step that occurs post BWA etc.