Resolve VEP versions

[kpjonsson]

The latest VEP release is 96 (http://grch37.ensembl.org/info/docs/tools/vep/script/vep_download.html#new). However, the neoantigen.py script in https://github.com/taylor-lab/neoantigen-dev does currently not support this version and recommends 88. For the sake of streamlining things, we will implement annotation of the somatic MAF with VEP 88.

The issue to resolve is that the HGVSc notation has changed between VEP versions and this column is parsed by neoantigen.py. In order to be able to use VEP 96 in the pipeline, someone needs to either: 1) Make neoantigen.py compatible with newer VEP versions, or 2) Write some script that can be run on a MAF to convert the HGVSc notation to the older format.

[evanbiederstedt]

cf Slack discussion on 13 June 2019: Chai @cband, Allan, KPJ, Evan

The line in question was here: https://github.com/taylor-lab/neoantigen-dev/blob/5a82501f357f1bdbecda22ebfe8960bbde1188f1/neoantigen.py#L686

I think we agreed to run maf2maf on the MAF using the vcf2maf container, using vep v88.

The only annoying thing about this to me is that v88 is from March 2017. http://www.ensembl.info/category/01-release/

Any future methods sections would need to say "we used this old version of vep88"

It looks like there were additions in v91, v94, v95 with updates for the HG assemblies. Maybe we should update...

The caveat is that Ensembl updates every 3 months, and no one has the bandwidth to keep making updates to our code.

[evanbiederstedt]

I think we're just going to bite the bullet here, and use v88 for everything

There will come a point when we get sufficiently annoyed we will updated the Ensembl version. We just need to make sure (A) the vcf2maf code has been updated and (B) Chai's python scripts parse the strings correctly

[kpjonsson]

See #353. Closing this for now.