Normally, I check these plots to see where (bp) to truncate in the DADA2 denoising step. Due to the binned quality scores of NextSeq, it is not so easy anymore to interpret where to truncate. (as compared to MiSeq)
Yes, DADA2 supports binned quality scores.
(And, there may be room for improvement. Both are true.)
Just like MiSeq, I usually run DADA2 10 times with different settings and pick the one that gets the most reads to merge. Guess and check works well enough.
If you want to post the stats from some of your 10 DADA2 runs, we can take a look.
Running multiple times and guessing sounds like a pragmatic way to deal with the binned scores, though feels a bit arbitrary. But I guess with this binned scores it's difficult to think of something else...
We are trying to maximize read retention, and I'm using guess and check instead of a more rigorous parameter search.
Trim too long, then fewer reads will pass the filter.
Trim too short, then more reads will pass the filter but some reads cannot merge.
Trim just right, numbers go up
It's not arbitrary; I'm selecting settings to maximize one of the columns from DADA2 denoising stats.
Alright thanks! I'll continue now with the analysis (had to work on a paper last week), but If I have questions regarding my specific results I'll post it here!
