Telling Heterozygous Mutations From Tcga Maf Format?

Question

[user]

Is it possible to tell heterozygous/homozygous mutations from the TCGA maf files from exome sequencing? It seems to me that the answer is no since there's no way to resolve the chromosome identity from exome-seq. Is that true?

Answer 1

[Chris Miller]

Some MAFs contain readcount and allele fraction information, which can help. Even with this info, it can be difficult, as amaro points out.

Say, for example, you see a variant at 70% percent frequency. This could be:

a) a het mutation with a 1x copy number amplification in a pure tumor (near 66%)
b) a homozygous mutation in an impure tumor (~30% normal admixture)
c) a het variant at 50% with low depth (sampling error)
d) some combination of the above.

Even if you see a variant at 100%, you have to consider whether it's a double hit somatic mutation, a copy number loss, or copy number neutral loss of heterozygosity (or a germline event that was falsely called somatic).

Answer 2

[amaro]

Being able to tell if a somatic mutation is heterozygous or homozygous from exome data is complicated and can't be determined just from a maf file or the read counts at a site. You need to know the copy number at the locus, the purity of the sample etc. to be able to determine this kind of information. However, you could use the absolute purity/ploidy calls + snp array data from the TCGA data portal to get cancer cell fraction or multiplicity calls for each of the mutations this might help you achieve what you are looking for.