DADA2 denoising filtering/merging step removing 60-70% of reads

Hello!
First of all, thank you for providing all the details and structuring your post. It saves a lot of time!

Based on your qzv files, your samples collected most of the issues that could decrease the number of output sequences:

  • filtering (low scores)
  • chimeras
  • merging

Considering that you already removed the primers and discarded sequences without them, I can suggest:

  • using 260 and 230 trunc. values for F and R reads
  • decreasing minimum overlap to 6
  • increasing max_ee values for F and R
  • setting --p-min-fold-parent-over-abundance to, for example, 8 (not more than 16)

and see how it affects your stats.

Best,