DADA2 Error Rate Plots - is this okay?

[suminayeoh]

Hello,
Thanks for having this dada2 community - I would like to be enlightened with my question please! Very new to dada2.

All my error rate plots for both 16s and ITS (F and R) came back with a similar pattern below:

I am aware good error plot should show that black line (estimated error rate) as good fit to the points (observed rates), and the rates drops as quality increases. But for mine, some plots show a dip at quality score around 30. I'm not quite sure but I would think these are harmless observations overall to the quality, since the points mostly follow a downward line? And I reckoned it is a harmless dip due to just the final point going upward. Butk keen to hear if anyone has any opinion about this!

This whole process is truly computationally intensive so I wonder should i re-run by adjusting the nbases or is this good enough to proceed? I have tried to run the subsequent processes - dereplication, sample inference and paired-end merging in a loop and it takes up to two days for each of my dataset ( i have four in total), so any advice would be really appreciated!

Many thanks,
SM

[AngusBall]

Hi SM,

I'm not a dada2 dev, but I've had this issue when running my nextseq data (or any data with binned quality scores e.g. novaseq). I'd look at issue #1307 as it helped fix my data when I had this issue.

or specifically this comment since the issue is chock a block full of information

goodluck!