Should I include ambiguously mapped reads in the realignment around InDels?

Question

[stu111538]

Hi,

I am wondering whether ambiguously mapped reads (=not uniquely mapped reads, mapping quality 0) do play any role when I perform realignment around InDels (as for example using GATK IndelRealigner or, as in my case, BisSNP BisulfiteIndelRealigner). I performed the alignment of my RRBS data using Bismark/Bowtie2 and excluded all ambiguously aligned reads from my bam file. However, could realignment around InDels do anything with those reads, so that they maybe are unambiguously aligned after the realignment?

Answer 1

[Medhat]

First:

As announced in the GATK v3.6 highlights, variant calling workflows that use HaplotypeCaller or MuTect2 now omit indel realignment.

https://software.broadinstitute.org/gatk/blog?id=7847

Remember that;

These Read Filters are automatically applied to the data by the Engine before processing by RealignerTargetCreator.

MappingQualityZeroFilter
MalformedReadFilter
BadCigarFilter
BadMateFilter
UnmappedReadFilter
NotPrimaryAlignmentFilter
Platform454Filter
FailsVendorQualityCheckFilter
DuplicateReadFilter
MappingQualityUnavailableFilter
  

Answer
you do not need to remove them.

more:

https://gatkforums.broadinstitute.org/gatk/discussion/7067/ahoy-mates-be-ye-unmapped

[stu111538]

in the second part, did I understand you correctly that when I run RealignerTargetCreator, this software starts with those filters you listed and after it does the actual target creation?

Thanks!

[Medhat]

exactly so as I answered no need to interfere with the main workflow of GATK, good luck :)