Method of Checking for Mutation Patterns

Question

[L. A. Liggett]

I am trying to think of an approach to understand mutation biases in my data if they exist, but I can't think of a good method. The idea is that I see particular base change biases in my genome sequencing data which match similar published data for example C mutates to T more often than C mutates to A. It is relatively straightforward to just look at the output vcf file and get this information.

However, now I would like to look for more complex patterns. For instance perhaps C often changes to T but the majority of the time this is only in the context of ACA -> ATA, because that the surrounding bases influence the error rate. Similarly we could imagine that any number of surrounding bases might influence the error rate such that perhaps AAACAAA -> AAATAAA is the most prevalent C -> T change.

So I am looking for some guidance or suggestions on how to proceed with an analysis like this. I know some labs have performed and published this type of data but I can't think of how to do it myself.

Answer 1

[igor]

I don't know if there is already a tool that specifically does this. Since you already have a list of variants, you can use something like bedtools getfasta to retrieve the surrounding sequence for each one. That gives you the genomic context that you would then have to summarize.

[L. A. Liggett]

Oh cool I wasn't aware of bedtools getfasta. That is helpful.

Answer 2

[Charles Yin]

You may try NNMF (Non-negative matrix factorization) to analyze the mutation signature of the mutation patterns. The signatures are based on k-mer pattern of the mutations. The following is good reference (Alexandrov, L. B., Nik-Zainal, S., Wedge, D. C., Campbell, P. J., & Stratton, M. R. (2013). Deciphering signatures of mutational processes operative in human cancer. Cell reports, 3(1), 246-259.) link to the paper

[L. A. Liggett]

You know this was one of the papers i was remembering, but i didn't remember that they linked to an explanation of how to run the analysis; I see that now, hopefully this will help me solve my problem. Thanks.