I have been attempting to pick OTUs with USEARCH in Qiime. I was suggested to split the jobs into batches of the groups (healthy and diseased) that I want to look into instead of running all the groups in one single run. This would be so that the machine won't be running for a long time. Would this splitting up the jobs affect how USEARCH searches for chimeras significantly?
For chimera having one would be fine. However, the run time would depend on what type you go for: de novo, or reference based.
USEARCH might be annoyed if file size is big, And you've 32 bits' distribution.
Also, at clustering step, you might want to have them one. Because that could/might/would be one of your goals to know how taxa are spread across case, and control. Clustering would bunch same OTUs as one, and on that output you'd do classification.
It might be worth getting in touch with Robert Edgar who wrote USEARCH for more technicalities. He is pretty quick in replying.
