Increase speed of classify-sklearn on HPC


I’m running the taxonomic classification, using a trained classifier, with the following command line:

qiime feature-classifier classify-sklearn
–i-classifier silva-132-99-v3v4-classifier.qza
–i-reads rep-seqs.qza
–o-classification taxonomy_silva-132-99-nb2.qza

I’m working using an HPC (multiple node, multiple CPUs, parallel jobs), so every time I submit a job, using SLURM, I can specify number of nodes requested, CPUs, etc…

For feature-classifier classify-sklearn, there are some options as:

So, I was wondering what are the best options to set up a faster job using the HPC? How should I define those options according to the HPC?

Thanks in advance

Use --p-n-jobs to parallelize this action. It will run much faster (though require more memory)

Only request as many jobs as you have available. Discuss with your HPC administrator to make sure you are requesting cores from the HPC in the appropriate manner.

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.