Q2-ITSxpress: A tutorial on a QIIME 2 plugin to trim ITS sequences

[benjjneb (Ben Callahan)]

@Adam_Rivers

I think that Dada2 has the ability to learn error rates from single ended data as well as paired end data now, but I don’t know the method. @benjjneb can you provide guidance on this? How are error profiles calculated in denoise_single ? Will merged scores interfere with the error rate estimation procedures?

Merging by BBMerge does change the quality scores since most positions are verified by two reads. The score change is shown for one ITS1 sample here:

In short, DADA2 does not recommend processing pre-merged reads, because the different regions of pre-merged reads (Forward-only, overlapping, reverse-only) have different relationships between the assigned quality scores and the error rates, which can lead to false-positive ASV inference. You can see more discussion of this here: Fungal ITS pre-processing suggestion · Issue #327 · benjjneb/dada2 · GitHub

This is a bit annoying, and we would like to support pre-merged reads, but when we evaluated this possibility again recently we were not able to attain the same accuracy on such reads as we could in our recommended merge-later workflow.