hello everyone,
I am currently using galaxy so as to analyze metagenome and I could not make SILVA 138 (99% OTUs, full-length sequences) work,the error showed > “This plugin encountered an error: The scikit-learn version (0.24.1) used to generate this artifact does not match the current version (1.4.2). Please retrain your classifier…” so someone told me to switch to galaxy eu because it could work there but there was the same error,I tried dowloading * * silva-138-99-seqs.qza
silva-138-99-tax.qza
and to make a classifier by myself with the tool called qiime feature-classifier fit-classifier-naive-bayes` and the error showed
“This job was terminated because it used more memory than it was allocated.”
what options do I have ? thank you so much!
But I do have a few thoughts. It appears that galaxy eu uses QIIME 2 version 2024.10. So you'd have to make use of sequence and taxonomy files from here to run qiime feature-classifier fit-classifier-naive-bayes ... This should avoid the scikit-lern version issue you are running into.
I just uploaded the files I referenced above, to galaxy eu, and I am now running qiime feature-classifier fit-classifier-naive-bayes .... I've not run into any errors yet. Though I suspect that I may run into the memory error you mentioned. I am not sure how much memory galaxy eu allows, but its probably not enough for constructing a classifier for the full-length Silva SSU database.
Assuming you are using the V4 region, I'd suggest, making use of the 515F/806R sequence and taxonomy files. They will require far less storage space and memory, and might help you avoid the memory issue.
If you are making use of another variable region, then I'd suggest following parts of this RESCRIPt SILVA tutorial. One quick way to make an amplicon-specific region is to download the full-length files I linked to above, and skip down to the part that describes making an amplicon specific classifier. Then upload your new files / classifier to galaxy eu.
