Mutect2 panel of normals design

Answered

Follow

Jagadheshwar Balan

• March 08, 2021 14:45

Hi, I am seeing recurrent technical artifacts and germline mutations in my set of tumor samples. I do not have a matched normal for the tumor sample to be able to do a robust somatic variant call. To help clean up the artifacts, I am looking to create a panel of normals (PoN) for somatic variant calling using Mutect2.

 

Related to that I have a few questions:

1. I realize that the minimum recommended number of samples is 40 according to the document for Panel of Normals by GATK team. Do the 40 samples need to be of same cancer type? Is there any heterogeneity okay in the panel of normals. For instance, can we have 20 endometrial cancer, 20 colorectal cancer, 20 brain tumor samples in our panel of normal? (Note that we will have sample library prep, sequencing protocols for these, but just cancer type might differ)
2. The document recommends blood as the typical sample type. Is FFPE acceptable? Our assay will be using solid tumors and samples are typically FFPE type here.
3. The document also recommends healthy/young samples, and this might not always be the case in solid tumor samples. I am wondering if this is okay. In some cases, the normals are normal tissue cuts right next to the tumor, are there specific red flags about using such samples?

Please let me know. Thanks in advance for your help and looking forward to hearing from the GATK team.

0

• 

  Genevieve Brandt (she/her)

  • March 10, 2021 04:45

  Hi,

  The GATK support team is focused on resolving questions about GATK tool-specific errors and abnormal results from the tools. For all other questions, such as this one, we are building a backlog to work through when we have the capacity.

  Please continue to post your questions because we will be mining them for improvements to documentation, resources, and tools.

  We cannot guarantee a reply, however, we ask other community members to help out if you know the answer.

  For context, check out our support policy.

  0

  Comment actions Permalink
• 

  David Benjamin

  • March 12, 2021 17:27

  Before answering your questions, we need to point out a few things:

  1. The panel of normals' role is only to clean technical artifacts, not remove germline mutations, which is handled by the germline resource and FilterMutectCalls.
  2. Since you don't have a matched normal, even with strict germline filtering you will end up with a very large number of germline false positives from rare (ie absent from gnomAD and other resources) germline events.  Any given genome will have 30,000 of these or so, and the only way around it is to have a low-purity or low-allele fraction tumor in order to distinguish somatic variants from germline hets.
  3. In most cases using the public panels from our google bucket is better than making your own.

  Now for your questions.

  1. The cancer type is irrelevant since technical artifacts are independent of the biological nature of the sample.
  2. FFPE is okay but I expect it to lead to a slight  (less than 1% if I had to guess) loss of sensitivity even when calling on FFPE samples.
  3. Don't worry about the age of the samples.  The concern is that non-cancerous somatic variation in older samples would lead to real variants getting into the panel, but this doesn't happen enough to be a serious issue.  Normal tissue next to the tumor will be a serious problem if a significant amount of tumor cells contaminate the normal, but not otherwise.
  1

  Comment actions Permalink

Please sign in to leave a comment.