Evaluation of a QIIME2 Classifier for Mitochondrial Cytochrome C Oxidase subunit 1

Hi all,

I’ve trained a taxonomic classifier in QIIME2 that has been trained on MT-COI sequences extracted from the Barcode of Life Database. Unfortunately, I have not found a way to evaluate my classifier to ensure that my results are accurate. My concern is that reviewers will be skeptical of this approach (as they should be). I’ve seen there is a program, TAX CREdiT, that can be used to evaluate classifier accuracy with mock communities, but there doesn’t seem to be a mock community available for COI sequences. Do you have any suggestions for assessing the accuracy of my classifier to ensure that my results are accurate?

Also, because no one in my field (to my knowledge) has used a QIIME2 classifier for assigning taxonomy, how would I best defend the classifier as an equal (if not, better) option for taxonomic classification, other than it is less time consuming and removes the possibility for human error?

I appreciate your time.

Thanks,

Brandon Hoenig

3 Likes

Hi @Brandon_Hoenig,

tax-credit also has methods for evaluating classification accuracy using cross validation — see the “cross-validated” section here. No mock community required :smile:

What are the gold standard methods for classification? Something like BLAST? You can compare multiple classifiers using tax-credit to see how QIIME 2 compares to other commonly used methods.

I hope that helps!

1 Like

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