Understanding Qiime2 Workflow

I’m analyzing samples derived from mouse stool. This is my first time using qiime. I have imported my pre-demultiplexed paired-end fastq files and run dada2 denoising on them. From the overview page (https://docs.qiime2.org/2019.7/tutorials/overview/#clustering), it states that clustering must be preceded by quality filtering and dereplication and followed by chimera and OTU filtering. The flowcharts show dereplication as producing a FeatureTable[Frequency], which is also the output from dada2, and the Demultiplexing flowchart shows data flowing through dada2 denoising OR through quality and dereplication modules. My understanding is that dada2 does quality and chimera filtering, so is dada2 denoising an alternative to OTU clustering, or are the filtering steps not required before OTU clustering if dada2 is used? Does dada2 do dereplication as well? If I’ve run the denoising, can I directly run OTU clustering on the output?



  1. dada2 is a better alternative to OTU clustering. It dereplicates, denoises (corrects sequence errors if a sequence is not too error-riddled), and then performs chimera checking.
  2. OTU clustering is NOT needed after using dada2 (since dada2 is a more sophisticated method of quality filtering and dereplicating your sequences), but it would be possible to run dada2 and then OTU cluster those outputs.

I’d recommend going with option 1 :wink:


If you so choose!

I hope that helps!


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