In the resources I have seen so far, the threshold is set when the first or second quartiles pass a quality score between 20 and 30. But with my data, it seems that this cutoff range seems irrelevant for truncating my data (especially reverse reads).
Does someone have an idea about this?
Thanks for your help in advance.
Yes, I agree. If the quality slowly declines, choosing the place it does as your truncation spot makes sense. But in your reverse R2 read, the quality 'bounces' up and down so there is no one clear place to trim.
Instead, try running DADA2 with a couple of trim settings. Pick the combination of settings that allow the maximum number of your reads to join. This 'guess and check' method often works well for me when my R2 read has variable quality.
Thanks for your answer!
I don't have access to a powerful computer at the moment, so running DADA2 on my dataset is quite long (several hours). Hence, I'll try your clever method soon when I can access better resources.
Thank you again for your time.