Unexpected run time, Trinity/salmon

[peterdfields]

Hi,

I've noticed that for some datasets the TransPi (conda controlled) seems to sort of get stuck in the Trinity assembly phase. Basically the pipeline stays in the salmon index phase. When I run the Trinity step (pulling the specific Trinity installation from conda matching that which is created as part of the pipeline) independent of the larger TransPi NextFlow recipe it completes successfully relatively quickly compared to how long I've been waiting for the TransPi pipeline to finish. Do you have any ideas about what might be going wrong?

[sevragorgia]

Hi Peter,

this is a trinity problem I am not sure how to fix. It seems trinity gets caught in the salmon step and creates huge output files without ever completing the run.

I have a fix but will need to check my Transpi.nf file to post it. Briefly, the Salmon step in Trinity is there to filter some missassemblies but according to the developers, it doesn't impact the assembly much. So in Transpi is probably 100% unnecessary because of the evigene step.

I will try to post the patch here and see if we can merge to the master branch.

Sergio