TCGA both blood and tissue for germline calling

Question

[Tiffany Delhomme]

Hi all,

I am currently analyzing TCGA data in term of germline variant calling. And as expected, for some cases I got both blood and tissue sample for the normal. This information can easily be found in the corresponding TCGA barcode ( https://wiki.nci.nih.gov/display/TCGA/TCGA+barcode ).

My idea is to use this to remove possible false positive, i.e. if a variant is not found in both samples (blood and tissue), I would to remove it.

Does anyone see a problem I didn't think about? Do you think it is a better idea to keep all blood variant (better quality than tissue), and add a variant from tissue if the QUAL is much higher?

[Tiffany Delhomme]

Thanks for your answer Kevin! I think I would consider all blood samples, and take the tissue ones only if there is no blood material available.

[GenoMax]

Please use ADD REPLY/ADD COMMENT when responding to existing posts to keep threads logically organized.

This should have gone under @Kevin's answer.

Answer 1

[Kevin Blighe]

Hi Tiffany,

My preference would always be to trust the blood normal over the tissue normal, i.e., give the blood normal the priority. The blood sample would have been extracted from leukocytes/lymphocytes from the blood buffy coat layer, whilst the tissue 'normal' would have been extracted from tissue surrounding the primary tumour, tissue of which has been shown to harbour neoplastic cells and somatic mutations related to the primary tumour.

Thus, the blood normal is a greater representation of germline DNA. Technically, therefore, everything that's encountered in the blood normal should also be found in the tissue normal, but not vice versa.

You neither want to filter out germline variants that increase risk of cancer, like those in BRCA1, BRCA2, the Fanconi Anaemia gene group, or the ERCC repair pathway genes.

Kevin

[Chris Miller]

Some caveats to your generally correct statement:

1) In heme cancers (i.e. AML), the skin is the normal.

2) It depends on what kind of caller you're using. Clonal hematopoesis is common in the elderly, meaning that somatic variants can be found at fairly high frequencies (up to ~25% VAF in otherwise healthy individuals). If your germline caller picks these up, you'll potentially be dealing with some false positives. The OP's idea of requiring presence in both blood and tissue would probably ameliorate this.

That said, I think you generally hit the nail on the head!

[Kevin Blighe]

Thanks for adding your thoughts Chris. Great points 1) and 2)

Kevin