Best practice for handling multi mapping quality

Question

[clear.choi]

We have data from the MiSeq and HiSeq machine and transferred to fastq file. We need to find SNP position and faced one issue with mapping quality.

We realized that "GATK4" HaplotypeCaller tool trimmed that multi mapping reads.

We want to use those sequences also.

Mapping quality is not actually equal to Zero. It is 19 but still "GATK4" trimmed that sequence.

Also in the sequence-tagged in the samfile It shows below. XA:Z:chr16,-55769830,203M,3;

So It looks like although mapping quality 19, It is not enough to score to find the snp position.

Does someone know how to handle in this scenario?

Thank you so much.

Answer 1

[Brice Sarver]

GATK excludes these by default; a quick search brings up quite a few questions on their forums about this. I can list a few things that may help point you in the right direction, but you may want to hop back to the mapping step and ask yourself what you should do when a read maps with equal probability to two different genomic locations and why that may be the case.

Assuming you're actually getting mapping qualities, based on your example above, what you want to look at is the -mmq option of HaplotypeCaller for 3.8. The default appears to be 20. For GATK4, the filters are less clear, but check out MappingQualityReadFilter and this post.

[clear.choi]

Thank you so much! There are two base mismatches and I think It should be able to handle by those two positions. I will go back to check mapping quality and other options.. I used GATK4 and I cannot adjust default Mapping quality read filter value.