detecting mutations in cancer samples belonging to the same cancer type

Question

[lait]

Hi,

I have tumor samples (wes) gathered from different patients. All the tumors belong to the same cancer type. I would like to check the mutations in those samples, if they affect the same genes, is there a correltaion, similar mutation pattern ... etc

My question is, how can I detect the mutations in the tumor samples? Those are not germline samples, so I cannot use germline piplines. Moreover, I do not have the normal samples in order to apply the somatic pipelines.

Could anyone guide me please on what to do in this case? Many thanks in advance

p.s. I understand that your answers are for educational purposes and you hold no responsibility. I have read such disclaimers in other posts.

Answer 1

[morovatunc]

There are mutation callers for this purpose. Depending on your sample you may want to use more than one or taking the intersections. (Algorithms have pros and cons for different sample types).

https://github.com/Illumina/strelka http://dkoboldt.github.io/varscan/ https://software.broadinstitute.org/gatk/ and more..

Also there is https://github.com/chapmanb/bcbio-nextgen that calls mutations with multiple callers and obtains their consensus throughout the pipeline.

EDIT: I just put some of the mutation callers that I have seen recently in the papers. For the best result, i suggest you to read a recent review paper and do your evaluation based on that.

[lait]

But those callers require either germline samples, or tumor-normal samples for the germline callers and somatics callers respectively, and what I have is only tumor samples that I would like analyse.

[poisonAlien]

If you have only tumor samples, no matter how much of a sophisticated algorithm you use - there is always an uncertainty associated with the somatic status of an identified variant. Having said that, MuTect can still call somatic variants without gremline controls. But again you will have to be careful and filter possible gremline SNPs against dbSNP, ExAC, etc.

[lait]

thank you a lot for the clarification. In this regard, what do you think of this tool named ISOWN?

[poisonAlien]

Its using Mutect2 under the hood for variant calling and doing LOT of filtering to remove possible FPs and gremline variants - which is GOOD. I think you should try it!