Mutation calling for pediatric cancer data

Question

[Gene_MMP8]

I have recently gained access to the St Jude children's hospital database with more than 700 paired tumour/germline samples for common and rare pediatric cancers, Up to now, I have called only somatic variants from tumour-normal matched samples using the GATK pipeline. Being new to the field, I am confused regarding how to call the variants using tumour /germline samples. I understand that germline mutations are the ones that you inherit from your parents. So, can you guide me through the entire process calling somatic mutations from paired germline/tumour samples (.bam format)?

[German.M.Demidov]

Don't forget that not only short variants shape the tumor, but also structural variants in general and copy number variants in particular - not sure if GATK calls them by default, but there are modules in gatk that do so

Answer 1

[ATpoint]

Tumor-normal and tumor-germline is the same, or am I getting your question wrong?

A so-called normal sample is a non-disease sample, e.g. from peripheral blood if the disease is e.g. lung cancer, or any soft tissue if the disease is e.g. leukemia, that is supposed to represent your germline genotype which, as you say, is what you inherited via the DNA from oocyte and sperm of your parents. In contrast, the somatic variants are those acquired n that specific tissue you took the sample from. Therefore, what you ask is what you apparently already did via GATK. Maybe I am getting things wrong, in that case please elaborate.

[Gene_MMP8]

Got it. So then it's just a different terminology then. Thanks for clarifying that. Now I can just use the GATK pipeline to call somatic variants, right? I guess I was confusing between germline variants and germline samples.

[ATpoint]

Yes, based on what you describe I think it comes down to terminology and you can call variants as you did before.

[Gene_MMP8]

Thanks for your help!

[Gene_MMP8]

This is an extension of the question that I asked previously. So I have discovered that from the same database that has tumor-germline bam files, I can get tumor-normal vcfs too. So I have tumor1.vcf and germline1.vcf. Similarly, tumor2.vcf and germline2.vcf. This means that each bam file has been aligned against their respective reference genome and the variants have been noted. Now I want to produce one single maf file for driver gene identification using MutSigCV. How to do so? I thought about merging all the tumor vcfs into one big TUMOR.vcf and normals into NORMAL.vcf. Now how do I extract somatic variants and produce one single maf file for further analysis?