Closed tsp-kucbd closed 1 year ago
You must have an old version of macsyfinder as this has been fixed last year : https://github.com/gem-pasteur/macsyfinder/issues/54
Also it's weird though, because --cpu 0 in hmmer does not tell hmmer to use all cpus (unlike macsyfinder), so your job might be killed for another reason. Anyway, you should update macsyfinder.
When restricting the amount of cpus with -w / --workers, that information is not passed on to hmm search which has a hard coded "--cpu 0" and will happily try to take over all cpus on a node and, in our case, get killed by the scheduling system.
in the profile file:
defensefinder/lib/python3.9/site-packages/macsypy/profile.py
Line 172 - 173the workers variable should get passed to the Profile class on line 74