search

mskcc / tempo

CCS research pipeline to process WES and WGS TN pairs
https://cmotempo.netlify.com/
12 stars 5 forks source link

HLA Polysolver #90

Closed evanbiederstedt closed 5 years ago

evanbiederstedt commented 5 years ago

HLA typing: POLYSOLVER (aniroula/polysolver)

Alternative: Optiype (zhanglab18/optitype) (https://github.com/FRED-2/OptiType) Alternative: HLA*PRG:LA, Dilthey---https://www.biorxiv.org/content/10.1101/453555v1

evanbiederstedt commented 5 years ago

We can do this on the raw MAF, and it will work.

allanbolipata commented 5 years ago

We can do this on the raw MAF, and it will work.

HLA Polysolver doesn't seem to take MAF as input.

allanbolipata commented 5 years ago

I'm running into random errors for HLA Polysolver. It's written in bash and the logs don't seem to say what the errors are, so I am gonna switch it for now and try using Optitype.

evanbiederstedt commented 5 years ago

We can do this on the raw MAF, and it will work.

BAM, not MAF. Apologies. The normal BAM alone should work, as the HLA from the normal should give us all the information we need for immunotherapeutic decisions/research/etc.

We'll edit the pipeline chart.

evanbiederstedt commented 5 years ago

I'm running into random errors for HLA Polysolver. It's written in bash and the logs don't seem to say what the errors are, so I am gonna switch it for now and try using Optitype.

I think the bummer here is that people seem to really like this tool. I'm happy to shelf it now, but I'm pretty sure we will be asked to get POLYSOLVER to work in a pipeline. Let's maybe open a new issue.

kpjonsson commented 5 years ago

@allanbolipata How are you running it?

allanbolipata commented 5 years ago

Let's maybe open a new issue.

Probably don't need to open a new issue; we can keep it in here since this is the ticket for it.

@allanbolipata How are you running it?

See https://github.com/mskcc/vaporware/blob/feature/hla/somatic.nf#L584

The error could stem from me running a test file and it's just not spitting out the expected output, so the Nextflow process craps out. This is potentially a huge red flag since it makes it appear like it's not flexible enough to take just any bam as input.

But I know very little about this, so I could be totally wrong.