TCGA somatic mutation

Question

[lilingjoyo]

Hi, everyone.

The somatic mutation level2 data in TCGA contains 3 types:BI mutation calling, BI automated mutation calling, and BI curated mutation calling.

What's the difference among these types? Which is appropriate for me to get samples' somatic mutation information.

Answer 1

[Cyriac Kandoth]

At this link there's a bit of text about DCC archive versioning that might help explain what you're looking at. I'll summarize it for BI (Broad Institute) as follows:

DCC Archive Version Format: <serial index="">.<revision>.0
Serial Index = 0 indicates Auto-GSC generated MAF, 1 indicates AWG-curated MAF
Revision = Current data revision (archive version 1.n.0 based on archive version 0.n.0)

In general, I recommend using curated mutation calling for most analyses, unless - you deeply care about sensitivity, and are experienced with filtering through false-positive mutation calls. But in rare cases like long indels in FLT3 or BAP1 that the automated pipeline cannot capture, the curated calls may have more calls than those in the automated calling data. I also recommend reading this if you haven't already - Working with MAF files (Mutation Annotation Format) from the TCGA (The Cancer Genome Atlas)

[jackson.brian.c]

I am using the protected mutations (raw vcf files). For some cancer types, there is no BI curated, only BI automatic and BI mutation calling. In this instance, which should be used? What is the difference between BI automatic and BI mutation calling?