Cancer Exome-Seq data without matched Normal

[woodydon]

Dear all,

I am aware of that tumor exome-seq data without matched normal will have difficulty for distinguishing somatic and germline mutations. I aim to find "important" somatic mutations from the data. I know incorporating 1000 genome project would help a bit. However, there are still too many mutations. Is there a way to redeem this kind of suffering?

Thanks,
Woody

[woodydon]

Any suggestions?

[Richard Finney]

Compare with known somatic mutations databases like "cosmic".

Theoretically TCGA is supposed to produce known mutations; I don't know how reliable they are and how much is included in Cosmic.

If you match a known mutation, then ... maybe probably.

Novel mutation but not in dbSNP or 1000 genomes ... then maybe .... can't really be sure.
There are probably somatic mutations in dbSNP.

[woodydon]

Novel mutation but not in dbSNP or 1000 genomes ... then maybe .... can't really be sure.
There are probably somatic mutations in dbSNP.

Hi Richard,

I like your comments. My tumor samples are a rare type of cancer which is not included in TCGA's collection. I have tried to substrate dbSNP and 1000 genomes (GMAF > 0), and I still have 194 mutations from 171 genes. The same 194 mutations co-existed in all three tumor samples. Therefore, I guess that most of them are germline mutations. But, it is possible that a few of them are important somatic mutations (SNP or indels).

Thanks,
Woody

[Baseless]

Like Richard, I'd first go for COSMIC, grab their gene list and look who is in there. Then filter for GMAF >15% in general population and head for potential meaninful stuff (start/stop codons, missense variants predicted harmful by SIFT/Polyphen, indels,...). If you have access to clinical information, check the genotypes. Something peculiar here? Take a look at the genes located in those regions (if you have to find those, ENSEMBL/Biomart is your friend). If you have anything useful regarding the phenotype, maybe check with clinvar or phenomizer to refine your candidate genes.
And if everything fails, see if you can get a germline reference - remission sample, fibroblasts, oral swabs if necessary.