Classifier from SILVA data base

[1115 (Jonathan Yoou)]

Hi, all!

While I was training my own classifier with SILVA data base, I found a classifier already trained on SILVA data base. I read several questions of classifier on QIIME2 forum, most of replies say to use it for classifying. Here is my confusion point. I checked the capacity they occupy, and classifier on SILVA database occupies almost 3 times more than the other, I trained individually following instruction on the https://docs.qiime2.org/2018.6/tutorials/feature-classifier/

The thing is, sometimes when I use the classifier from SILVA works, and sometimes it doesn’t while the classifier I trained works. What is the difference between them?
My guess is the classifier from SILVA data base is trained with all, 16S , 18S and ITS and what I trained is trained with only 16S.
If so, is it possible to take more time when I use the classifier from SILVA data base?

Thank you!

[Nicholas_Bokulich]

@1115

My guess is the classifier from SILVA data base is trained with all, 16S , 18S and ITS and what I trained is trained with only 16S.

Yes, that is the difference.

@1115

If so, is it possible to take more time when I use the classifier from SILVA data base?

Yes, the classifier trained on the full database would be more time-consuming. Smaller database sizes (e.g., only 16S) and shorter reads (e.g., trimming your reads to to the amplicon you used using extract-reads) will be faster.

I hope that helps clarify!

[1115 (Jonathan Yoou)]

Thank you, @Nicholas_Bokulich!!! Your reply helps me a lot!!!