Merge feature tables before or after OTU generation from qiime dada2 output?

I have 18 batches of multiplexed samples that I need to analyse together. Should I merge the qiime2 dada2 generated feature tables before or after OTU clustering with qiime2 VSEARCH?

The theory of OTUs would lead me to say merge before as OTUs are emergent of a specific dataset?

Hi @Lamm-a,

If I understand your set up, and assuming each batch is using the exact same primers, you'll want your workflow to look something like this:

  1. For each batch: demultiplex then run dada2 (taking care that your output will all be of the exact same region, meaning you have to make sure you use the same trimming parameters for each batch; if using just your forward reads, your truncating parameters need to also be the same)
  2. Merge all resulting feature-tables
  3. Run VSEARCH to cluster OTUs on the merged table

And yes you are right that clustering is data-specific as your clustering may look slightly different if you were to run it on each batch, that is unless you are using closed-ref OTU picking because in that case the reference used for OTUs will be the same.


Hi @Mehrbod_Estaki,

That is my exact setup. Thanks for the response :slight_smile:

1 Like