Can I keep the classifier.qza from q2-feature-classifier traning for future using?

Dear All, I train the q2-feature-classifier with different db(gg and Silvia) and also for different primer sets. I generated the different classifier.qza files and I rename them like gg13_5_v1v3_classifier.qza, gg13_8_v3v4_classifier.qza, etc.

I am wondering can I use those files directly in the future as the reference for different sample types instead of re-training the feature-classifier again?

Thank you very much.

Hey @archie,

Absolutely! That’s why we provide some pre-trained classifiers, they are applicable to any other dataset so long as the trained data contains the primer-pair used. Sounds like you’ve got a nice library of primer-specific classifiers set up :slight_smile:

Thanks, ebolyen.

Btw, where can I download the pre-trained classifiers?

If I didn’ t train the classifiers with primer parameter, the classifiers should fit all kind of library, right?

Appreciate it.

Btw,is there parameters like classifierPerIdentity=0.97 for the feature-classifier ? Thanks


Just so long as they are the same marker gene and internal to the full-length gene (e.g., any 16S amplicons could be classified with a full-length 16S classifier)

Use the --help flag to get information on any method, or look here.

Depends what method in feature-classifier that you plan to use. The classify-sklearn method has no such parameter, as it is not alignment based. The classify-consensus-* methods do.

I hope that helps!

Thank you so much, Nicholas

1 Like

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