Is there a vcf file on the GRCh38 assembly with common cancer mutations I can download somewhere? Maybe from one of the big international cancer genomics consortia?

By common, I mean whichever mutations have been found recurrent in different types of cancer.

  • $\begingroup$ How would you define this? Any variant found in >1 cancer type? In how many patients? Do you need to exclude variants found in only one type of cancer then? Should these only be causative variants? What if a variant has been found in three different cancer patients but has absolutely no connection to their cancer? And yes, that is not only possible, but almost certain to happen since many variants are very common in the population. Would you want to limit this by variant frequency? Please edit your question and narrow it down. $\endgroup$
    – terdon
    Commented Jun 15, 2017 at 12:40
  • $\begingroup$ @terdon It sounds like you are assuming he is doing population genetics, which may or may not be the case. In cancer genetics, they usually only care about somatic variants, so that automatically excludes "variants are very common in the population". I guess what the question really needs is a clarification regarding germline versus somatic mutations. $\endgroup$
    – burger
    Commented Jun 18, 2017 at 17:12

3 Answers 3


Your question doesn't give enough information for a specific answer but this should do for a start.

  1. Get the VCF file describing all variants in Clinvar from NCBI:

    wget ftp://ftp.ncbi.nlm.nih.gov/pub/clinvar/vcf_GRCh38/clinvar.vcf.gz
  2. Extract any variants whose VCF line contains the word "cancer":

    zgrep -iE '^#|CLNDBN=[^;]*cancer' clinvar.vcf.gz > cancer.vcf

Now, you should probably filter that further based on variant frequency, type of cancer, etc etc but that should be a good start.


If you are looking for cancer mutations, the primary resource is COSMIC and they provide GRCh38 VCFs. The download page is here: http://cancer.sanger.ac.uk/cosmic/download

It'll be up to you how you define "common", but the VCF includes a lot of information you can use.


The trouble with looking for cancer-associated variants is that it can be difficult to tease out spurious effects (e.g. ethnicity) from causative variants. If you're interested in what genes are implicated in most types of cancer, the annotations for the NanoString panels are quite good:

The "Support documents" section of those pages has a link to an Excel file with gene lists and probe sequences. While these lists don't give information about variants that disrupt function, I'm struggling a bit to think through why it would be useful in advance to know that, unless the aim is to induce cancer through CRISPR or something like that.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.