search

tmrcpsu / bacseq

7 stars 1 forks source link

Installing BacSeq #1

Open alex-trist opened 11 months ago

alex-trist commented 11 months ago

Hi! I've been trying to install BacSeq, but haven't succeeded yet. The main issue is that conda takes forever to solve the environment, even at the first steps (MultiQC). Installing the programs separately (without conda) would affect BacSeq performance? To restart the installation process, is it enough to remove the conda environment and run the install.sh again? Thanks!

hazmup commented 11 months ago

Have you tried setting the channel priority setting of conda to flexible by conda config --set channel_priority true?

alex-trist commented 11 months ago

Have you tried setting the channel priority setting of conda to flexible by conda config --set channel_priority true?

Thanks @hazmup. That solved the MultiQC issue, but now other conflicts appeared.

I've tried installing python 3.9 in the environment nad works properly until the abricate step. But no annotations (prokka, eggnog, staramr) and no phylogenies are generated.

Any thoughts?

hazmup commented 11 months ago

Are you installing python 3.9 manually? The env specification calls for version 3.7.

Is the rest of the software installing properly? What do the checks show when you run source run_BacSeq.sh?

alex-trist commented 11 months ago

Yes, I installed python 3.9 manually, because with python 3.7 I was only able to install abricate v0.4 (which takes different arguments and comes with different databases than the ones mentioned in abricate.sh).
Running run_BacSeq.sh shows the paths for all programs; however, the pipeline itself crashes after abricate (no output for functional annotation, pangenome, phylogeny, etc.). I am inspecting the log for answers, however I recently upgraded conda to 23.10, and was thinking of giving the new libmamba solver a try with a fresh install of the entire environment. Captura desde 2023-11-22 08-20-40

hazmup commented 11 months ago

I probably had the same problem and modified the install.sh file to use mamba instead of conda to solve the environments, and it solved it. I am now having some trouble running bacseq, specifically at the eggnog-mapper step, so I will get back to you when I figure this out as well. In general, I am using mamba for everything and it's incredibly faster and just as reliable. Maybe bacseq should be modified to use this by default, and also to specify package versions explicitly.