Finding Somatic Mutations From Whole Exome

Question

[Angel]

Q1) How do I call and report somatic mutations after getting the vcf files as below?

samtools mpileup -u hg19.fa foo.sorted.bam | bcftools view - > foo.sorted.bcf

bcftools view food.sorted.bcf | vcfutils varFilter -D200 > foo.d200.flt.vcf

[Rlong]

Do you have a matched normal sample? Without a comparison, you won't be able to tell somatic from germline mutations.

[Larry_Parnell]

It is not likely that all cells whose DNA was sequenced originated from the same clonal population. It is much more likely, especially for a clinical sample, that they've sequenced a mixture of cells of different origins, and hence different genomes. A matched normal is helpful, but may not be necessary.

Answer 1

[Sean Davis]

Take a look at the -T pair option of bcftools. A non-comprehensive list of somatic variant callers includes varscan, somatic sniper, joint-snv-mix, and mutect (licensed from the broad). I suspect that many folks use a custom script for this application, though.

[Angel]

Thanks so much. Sorry again for my delayed response. I will be trying some of these today.

Answer 2

[Dan Koboldt]

For exome sequencing of tumor-normal pairs, we use VarScan and SomaticSniper.

[Angel]

Thanks very much and I will begin with varscan.

[Sean Davis]

VarScan2 is out:

http://genome.cshlp.org/cgi/content/abstract/gr.129684.111v1