How To Differentiate Driver Mutations From Common Mutations And Snps In Cancer Rna-Seq Data

Question

[monukmr98]

Hi all,

I am working on RNA-seq data ( our own experimental data) for some cancer types of human. I want to know whether the genes found/mapped in the raw data have mutations significantly different from the normal mutations found in the cells. (i.e. distinguish somatic mutations from germline mutations). Also I want to negate significant driver mutations from the SNPs. So what packages are available for this task (what is the protocol exactly) ?. Also, which databases I have to compare my data with to exclude the obvious mutations/variations in my patient samples?

Thanks

Regards

Kumardeep

Answer 1

[Sean Davis]

What is typically done is to compare tumor to matched normal to define somatic mutations. However, this will be challenging in RNA-seq data since many genes expressed in the tumor will not be expressed in the normal. Also, variable levels of expression contribute to difficulties in calling variants in RNA-seq. These two issues reduce the ability to successfully call somatic variants from matched tumor/normal RNA-seq.

As for databases to compare with, dbSNP, NHLBI ESP, several personal genomes, Complete Genomics datasets, and several others detailed here:

http://bioinformatics.oxfordjournals.org/content/early/2011/09/27/bioinformatics.btr540.full.pdf

You'll also want to be aware of the COSMIC database to see if variants have been observed in cancer before.

Tools like annovar, snpEff, SIFT, polyphen2, and others can help to determine the potential impact of individual variants on genes.