Suspicious TCGA tumor samples

Question

[jtull]

I am analyzing a set cancer sites from the TCGA and I came across some samples that have no copy number aberrations throughout the genome. I assume that's possible for some but for some cancer sites, the number of those just seems more than I'd expect. So, my question is, is there a way to check for possible errors in the TCGA, e.g. tumor-normal swap?

[dsull]

Hi! Can you provide us with the cancer type and the TCGA patient / sample IDs of a few of those samples so that the rest of the community can also look into this?

[jtull]

Sure, a couple of them for instance from the TCGA PRAD: TCGA-A6-5659-01B-04D-A274-01, TCGA-D5-6930-01A-11D-1923-01, TCGA-DM-A1HB-01A-21D-A182-01, TCGA-78-8648-01A-11D-2389-01, TCGA-86-A4P8-01A-11D-A24O-01, TCGA-L4-A4E6-01A-11D-A24C-01, TCGA-56-8623-01A-11D-2391-01, TCGA-85-A4PA-01A-11D-A256-01, TCGA-90-6837-01A-11D-1943-01, TCGA-3A-A9IJ-01A-11D-A396-01, TCGA-F2-6880-01A-11D-2153-01, TCGA-F2-7276-01A-11D-2153-01 Thank you.

[dsull]

These appear to be from TCGA COAD.

I checked their profiles briefly on firebrowse.org -- and I unfortunately don't have a perfect explanation for the lack of copy number alterations.

Here are a few ideas:

1. Perhaps check the tumor stage? The tumor could be at a very early stage where there is very little genomic instability.
2. Check for some tumor markers -- e.g. BIRC5 (survivin) mRNA expression (which is almost always higher in tumor than matched normal), p53 mutation, KRAS mutation, etc.
3. Check their mutational burden. For example, the TCGA-D5-6930 sample actually has a pretty high rate of non-synonymous mutations (especially compared to the normal samples) so I'm not convinced it's a "normal" tissue.

[jtull]

Yes, from COAD, not PRAD. Sorry about the confusion. Thanks for your suggestions.