Trained classifier output not found

[cielo]

I am training a V3/V4 classifier using SILVA 138 but there is no generated output.

I am using qiime2.2024.10. I downloaded the reference sequences and taxonomy from here:

Then, I extracted the reads for the V3/V4 classifier.

I tried to train the classifier using this code.

But the output is nowhere to be seen in the indicated folder.

Please help.

[cielo]

I saw in another post that it could be a RAM issue but I have 32GB RAM and the highest memory usage was only at 85%.

Classifier code is running without a result

Hey there, I have a problem to get output from this code: !qiime feature-classifier fit-classifier-naive-bayes --i-reference-reads bact_seqs_v1v2_clean.qza --i-reference-taxonomy bact_tax.qza --o-classifier bact_classifier_v1v2.qza sequence and taxonomie are cleaned and sequences can be found in taxonomie.fastq data...but this code produce no output data "bact_classifier_....qza"...i get no error message....everything looks fine, but its not working. I use wsl ubuntu environment and qii…

[gregcaporaso]

Hi @cielo, My first thought was that this failed (maybe due to memory) but that the failure is not being reported through Jupyter Notebook. It does look as though the job is completing successfully. Was there are green line of text printed to the notebook output similar to the one in the previous cell saying Saved...?

A couple of things to try.

First, are you certain that you're looking in the right directory in your explorer window? I can't tell if there is an other directory named Chap3 that might contain the results. Here's what I'm looking at:

A more definitive way to check that would be to insert a new cell in your jupyter notebook just below the one you share the screenshot of where you train the classifier to run the following:

!pwd
!ls Chap3/classifier

I would be running those to check that I'm working and looking in the directories that I think I am.

If that doesn't reveal the issue, I might try to run this from the command line directly, rather than through Jupyter Notebook, to see if you get additional failure reporting.

Finally, I would try running this on a system with more available memory - e.g., 64GB as a next step. I think that's the most likely reason for a failure. Even if this job was only using 85% memory, it's possible that the memory used for other processes (e.g., virtualization, the notebook server, etc) might be pushing you over the limit.

Hope this helps! Let us know if you need additional help.

[cielo]

Hi @gregcaporaso

There was no green line below saying Saved. I only have one folder named Chap3 and one classifier folder under that so I am sure that I am working at the right directory. I will try to lessen the memory used but I do hope there would be more insights on why this is happening.

[gregcaporaso]

Hi @cielo, I think the most likely explanation then is that the job is running out of memory, but that that information isn't being passed up to the Jupyter Notebook (probably b/c available memory has been exceeded). There are posts around about dealing with high memory requirements for classifier training - let us know if you need some additional help. Good luck!