Has anyone looked into how the new quality binning strategy Illumina is using on NovaSeq sequencers impacts denoising of 16S data? What are the best practices?

To decrease per-sample sequencing costs in some large studies, our data was produced in a NovaSeq instrument. When we started working through the datasets our quality plots looked very odd, which led us to identify the new quality binning as the cause. (thanks to some posts here!!)
The question is: are there any best practices for working with this data? We are concerned that using any algorithm that takes into account quality scores is no longer appropriate, however, we are at an impasse regarding what the best approach is.

Do you have any advice?

Did you already find this issue on the Dada2 github? https://github.com/benjjneb/dada2/issues/791

So, options are:

  1. Use Dada2 R package with modifications as described by the link.
  2. Use Dada2 R or q2-Dada2 as it is, since there are several reports with not significant variations between using Dada2 as it is (R or Qiime2) and modified version
  3. Use Deblur which is not affected by changes in quality scores bining.

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.