I’m wondering if it’s recommended that scrapping or trimming of reads using trim.seqs (using qaverage, qwindowaverage, etc) after shhh.flows is recommended. I noticed in the shhh.qual file that there are reads with quality scores below 100. In your Schloss SOP, you don’t actually do it. Trim.seqs was basically used to separate the barcodes. And since shhh.flows doesn’t produce the usual quality scores (instead the scores are based on a 100 point scale), what cut-offs should be specified in order to obtain certain expected error rates? I’m a little bit of a stickler about errors because I’m looking for intraspecific allelic differences instead of OTU’s.
No, we don’t do anything with the shhh.qual scores. As we understand from Chris Quince these values are somewhat “experimental”.
Sorry to resurrect this 5-year old post, since then, has there any use of the shhh.qual scores?