Best quality control step for vsearch OTU clustering

[reige012 (Alicia R)]

I'm trying to use the vsearch function for OTU picking. I know that this is not the general preferred method for QIIME users, but I would like to do some comparisons using 97% OTUs instead of ASVs. My trouble comes in the merging and "cleaning" of sequences before clustering with vsearch. I have two main questions:

1. Should I join the paired reads using vsearch join-ends before or after the quality control step none of the tutorials are very clear on this?

2. What exactly is the best method of quality control on sequences being set up for OTU clustering vs. ASV?

I know that normally dada2 deblur/denoise is the quality control AND clustering step and
I did see the tutorial here (Clustering sequences into OTUs using q2-vsearch — QIIME 2 2018.11.0 documentation) suggesting that the data need to be quality controlled, but no real suggestion on the best method.

My trouble is I cannot figure out if I can use dada2 sans deblur or denoise since these are both ASV clustering methods. Is it possible to use dada2 just for quality control? Or is there some other more preferable method that I am just missing somewhere hidden in the tutorials?

Thanks for your help.
Alicia Reigel

[Nicholas_Bokulich]

@reige012

• Should I join the paired reads using vsearch join-ends before or after the quality control step none of the tutorials are very clear on this?

before

@reige012

1. What exactly is the best method of quality control on sequences being set up for OTU clustering vs. ASV?

use q2-quality-filtering would be typical qiime1-style quality filtering based on per-base quality scores.

I hope that helps!