-
Notifications
You must be signed in to change notification settings - Fork 42
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add an implementation of Mutect #114
Comments
Would it make sense to incorporate ideas from Strelka as well? Specifically ability to do indels? MuTect is the best out there now for mutations though. This would be a killer feature!! |
@jstjohn that's definitely on the roadmap! We should have the local assembly front-end reintegrated soon, which should give us a better ability to resolve INDELs than Strelka. |
Is this patent application going to get in our way? I have some experience reimplementing Mutect, but then I found this: https://www.google.com/patents/WO2014036167A1?cl=en&dq=mutect+getz&hl=en&sa=X&ei=YBPLVJfoHsOtogT7zIGgBw&ved=0CB0Q6AEwAA |
Reviving this thread, based on conversations with @jstjohn. Things that need to be implemented to be feature complete: Read filters:
Likelihood model:
Variant filters:
I don't know the details about the patent issues. CCing @massie to comment. |
The statistical model looks nearly complete so far. I am starting to pull @tdanford's mutect code into a fork of your A few thoughts and questions so far:
|
Tests are progressing nicely. Maybe 50% done with testing/fixing the likelihood model. Found a few bugs that were a result of @tdanford getting tricked by Nature's poor equation formatting. When they meant Work is here if you are curious pre-PR: https://github.com/jstjohn/avocado/tree/mutect_work |
Just finished up some initial tests and opened the pr #167 to track progress. |
Quick status update on mutect filters that are functions of the alleles/reads supporting the mutation in the tumor (where 90% of the special sauce lies). Done/potentially needs testing:
TODO:
|
marking all complete, although more tests need to be written on everything other than the second, fourth, and final bullet points above. Basically the sum of mismatching bases needs to be properly calculated in the ReadExplorer code, the mate rescue status needs to be passed through, need to implement the end-cluster filtering based on the MAD/median, and the power-based strand bias filter. |
TODO:
Other than that the filters are all in place, and the likelihood model is doing what it should be doing. I modified the code slightly so it should do something reasonable if it encounters an indel and you want to try out using the Mutect algorithm on indels. That is disabled by default. |
See this paper: http://www.nature.com/nbt/journal/v31/n3/abs/nbt.2514.html
(and the supplementary information) for details.
The text was updated successfully, but these errors were encountered: