Genome Analysis Toolkit

Variant Discovery in High-Throughput Sequencing Data

GATK process banner

Need Help?

Search our documentation

Community Forum

Hi, How can we help?

Developed in the Data Sciences Platform at the Broad Institute, the toolkit offers a wide variety of tools with a primary focus on variant discovery and genotyping. Its powerful processing engine and high-performance computing features make it capable of taking on projects of any size. Learn more

Declared strand bias not observed in igv (only in local reassembly)


1 comment

  • Avatar
    Gökalp Çelik

    Hi Fran

    Mutect2 uses local reassembly to generate genotype calls therefore there are many parameters involving the decisions given within each local reassembly. 

    Looking at the bam views and your VCF line your variant is not only filtered due to Strand Bias but also due to clustered events nearby as they can be seen in the IGV image. Those clustered events are mostly due to low BQ basecalls therefore most of them get pruned within the local reassembly causing a strand bias for the reference nucleotide G. This in the end causes the whole variant to be filtered due to clustered events and strand bias. 

    Looking at the command line parameters you have you seem to have modified your 

    --max-events-in-region 4

    which is directly involved in the creation of the clustered events filter. Those reads look like there may be more clustered events therefore increasing this parameter may change the way this variant gets filtered in the local reassembly. 

    Also since those clustered base changes will be of low BQ using the parameter 

    --min-base-quality-score 20

    will get those bases pruned almost immediately. The default value of 10 is a much less strict than 20 and may help your reference base to be included with more balanced strand values. 

    Before the final one additional comment from our team is that although SB annotation in the FORMAT field is shared between HaplotypeCaller and Mutect2, unlike HaplotypeCaller Mutect2 does not use Fisher Exact Test for the strand bias filter but has its own model. 

    If you wish to get into more details about the local reassembly we do have documentation in the link down below. 

    Also if you wish to understand more about the filtering strategy 

    I hope these help.


    Comment actions Permalink

Please sign in to leave a comment.

Powered by Zendesk