q2-sidle running out of memory at reconstruct-counts

Library Plugin Support
5

Sure! I preprocessed the SILVA database with RESCRIPt. I pretty much followed this. More specifically, after getting SILVA data (NR99, version 138) with qiime rescript get-silva-data:

  1. Remove low-quality seqs with qiime rescript cull-seqs
  2. Dereplicate identical seqs with qiime rescript dereplicate
  3. Extract specific regions from ref-seqs for each amplicon based off the primers sequences with qiime feature-classifier extract-reads

After these steps I have a high-quality dereplicated SILVA database for each amplicon*. Running these preprocessed databases in parallel in qiime sidle prepare-extracted-region gives me the --i-kmer-maps referenced in the qiime sidle reconstruct-counts command.

*Because I am running other analyses using these very same preprocessed databases, I am confident they are fine.

I hope this clarifies the situation. Thank you for your help!

1 Like