unexpected variant calls by Mutect2 depending on ExtractIlluminaBarcodes --MAX_MISMATCHES

Answered

Follow

myourshaw

• July 08, 2021 22:47

If you are seeing an error, please provide(REQUIRED) :
a) GATK version used: Picard 2.25.5, GATK 4.2.0.0
b) Exact command used:
c) Entire error log:

If not an error, choose a category for your question(REQUIRED):
a)How do I set MAX_MISMATCHES and related ExtractIlluminaBarcodes parameters?

I'm running a solid tumor panel with Mutect2 in tumor-only mode with a PoN on data produced by Illumina Next Seq and NovoSeq. We are using IDT xgen dual UMI adapters. The barcodes have a total of 16 bases and a minimum Hammig distance of 6. In an investication of the impact of using in ExtractIlluminaBarcodes  --MAX_MISMATCHES of either 0 or 1 (the default) and other related parameters as default, we see surprising results. A number (~10 to 100+) of called variants for a sample with the 0-mismatch parameter are not called with the 1-mismatch parameter, and vice versa. This phenomenon does not appear to correlate with read depth or variant allele frequency. We are particularly concerned by the possibility that Mutect2 will have a false negative for a variant depending on the small difference in supporting reads between the two values on MAX_MISMATCHES. Under what circumstances might Mutect2 fail to call a variant that has, ~700 reads 35 of which are single base mutations, when MAX_MISMATCHES is 0, and yet call the variant with MAX_MISMATCHES as 1 and 710 reads, 36 mutations?

In addition to MAX_MISMATCHES, should we also use non-default values for MAX_NO_CALLS and MIN_MISMATCH_DELTA?

 

0

• 

  Pamela Bretscher

  • July 09, 2021 12:56

  Hi myourshaw,

  
  I am going to move your post into our Community Discussions -> General Discussion topic, as the somatic topic is for reporting bugs and issues with GATK.
  You can read more about our forum guidelines and the topics here: Forum Guidelines.

  
  Best,
  Pamela

  0

  Comment actions Permalink
• 

  Genevieve Brandt (she/her)

  • July 16, 2021 20:25

  Hi myourshaw,

  Could you give some specific examples of what you are seeing? More specific examples of a difference you see when MAX_MISMATCHES is 0 and 1 would help us problem solve to figure why the tools are behaving this way.

  Please also provide your complete commands.

  Best,

  Genevieve

  0

  Comment actions Permalink

Please sign in to leave a comment.