I want to use vsearch for a metabarcoding pipeline, I have used UPARSE from USEARCH before. This tool had a build in chimera checker.

My question is about this page: https://github.com/torognes/vsearch/wiki/VSEARCH-pipeline

Is it necessary or usefull to cluster first at 98% and than do the chimera check and after that clustering at 97%? And what exactly does the perl script? I am not familiair with perl. Sometimes I see chimera checking after clustering and sometimes before, what is the best way?

Before I saw this page I was planning to do (without that pre-clustering):

1. Dereplicate
2. Chimeracheck
3. Cluster or denoising

Also, if I pre-cluster on 98% using a denoising tool does not make any sense.

Thanks!