Difference between variants detected using gatk 4.1.8.1 and 4.1.9.0

Answered

Follow

Nefté Yañez

• August 31, 2022 06:30

I use Mutect2 -tumor only with PON generated with 30 samples; the origin of the samples has the following features:
-Panel of 143 genes 
-Average depth of 1200x
-sequencing system Hiseq 2500 for a 125 base paired-end
-sample of peripheral blood
Both versions (gatk 4.1.8.1 and 4.1.9.0) were tested but there is no match in the depth of the reported variants and IGV.
In addition to this, the parameter -- max - reads -per- alignment - start was disabled to see if there was a greater coincidence in the reported depths.
The results showed too much discrepancy in what was reported by Mutect2 and IGV; added to this the total of variants found between the versions doesn´t match.
I understand that there are differences between DP, AD, and IGV but my results seem spurious.

b) Exact command used:

gatk Mutect2 -R ucsc.hg19.fasta -I x.bam -tumor x.bam --germline-resource af-only-gnomad.raw.sites.hg19.vcf.gz --af-of-alleles-not-in-resource 0.0000025 -L xxx.interval_list --panel-of-normals PoN.vcf.gz -O xx.vcf.gz

and 

gatk Mutect2 -R ucsc.hg19.fasta -I x.bam -tumor x.bam --germline-resource af-only-gnomad.raw.sites.hg19.vcf.gz --af-of-alleles-not-in-resource 0.0000025 --max-reads-per-alignment-start 0 -L xxx.interval_list --panel-of-normals PoN.vcf.gz -O xx.vcf.gz-

-max read per alignment start 50 :

0

• 

  Anthony DiCi

  • September 06, 2022 17:09

  Hi Nefté Yañez,

  Thank you again for writing to the GATK forum! I appreciate you taking the time to repost in our support language. I hope that we can help you sort this out.

  When reviewing your inquiry, I noticed a couple of things. Firstly, both of the versions of GATK (4.1.8.1 & 4.1.9.0) that you used are out-of-date. Additionally, you mentioned that you ran Mutect2 with a Panel of Normals (PoN) composed of 30 samples. 

  1. We always recommend you use the latest release of GATK (4.2.6.1), as many bugs from previous versions are already corrected.
  2. We strongly recommend you use a PoN of at least 50 samples when running Mutect2.

  That said, I recommend you try upgrading to the latest version of GATK and running Mutect2 again with at least 50 samples in your PoN. If the discrepancy persists after implementing these changes, we will hopefully better understand both the origin of the problem and our best next steps.

  I’ve included the latest tool index for Mutect2 here and our technical documentation on creating PoNs here.

  I hope this helps! Please let us know if these suggestions lead you to success. If you require any further clarification or if other questions arise, please do not hesitate to reach out. 

  Best,

  Anthony

  0

  Comment actions Permalink
• 

  Anthony DiCi

  • September 13, 2022 17:44

  Hi Nefté Yañez,

  We haven’t heard from you in a while so we will be closing out your ticket in our system. If you still require assistance, you need only respond to this thread, and we’ll make you a follow-up ticket to pick up where we left off.

  Thank you again for contributing to our GATK forum!

  Best,
  Anthony

  0

  Comment actions Permalink

Please sign in to leave a comment.