How to find true mutations in tumor by removing the background from matched sample?

Question

[dodausp]

In other words, How to normalize tumor SNPs by using matched sample??

.

Hi, everyone!

So, I have a dataset of somatic mutations consisting of 3 samples for each individual, that being: blood, primary tumor and recurrent tumor. I would like to take advantage of the blood samples and use it as a reference for each individual. Some sort of "germline" if you will. Now, I am aware that blood samples might not be ideal, as germlines are. However, I thought that this would be a valid work-around in order to find true mutations in tumor.

However, I would truly like to have your expert feedback on this:

1. Would this be a valid way of doing it?
2. And in case it is, what pipeline/algorithms would you recommend to use? (preferably in R)

I am sorry for the seeminly simple question, however I couldn't find a thread exactly on that topic here. This was the closest I could find, but still not quite: A: Strategies to call variants from a cancer sample

Any light on this will be very much appreciated! (:

Answer 1

[bernatgel]

What you want is to call somatic variants and for that you should use a somatic variant caller, since the problem is actually quite more complex than removing the variants present in the blood sample.

There are many tools for that, for example:

• GATK Mutect2
• Strelka
• Varscan
• FreeBayes

and many others.

This review paper has a list of somatic variant callers that might be useful to you. In any case I would start with one of the tried-and-tested and widely used options suchs the ones listed above