I am trying to understand if there is anything wrong running OTU clustering (open) with a table that are ASVs. I don’t think this should be any different than using all sequences as the input, but maybe I’m missing something. I am asking because my resulting ‘OTU’ table has only ASVs and nothing from the reference database.
There’s nothing technically wrong with it, but at this point, it’s less recommended. The goal of ASVs is to give you higher resolution than OTUs which you lose i you chose to cluster the sequences (even if you’re clustering at 99%).
Its a little bit weird that your OTUs aren’t hitting the reference, but that may depend on your reference, your algorithm, etc.
Are you running paired end Dada2? Because DADA2 will pair the ends of the sequence. If you don’t have overlap, I think the recommendation is to discard the reverse read and cut your losses. If you are scaffolding with the reference, you need to use closed reference clustering alone, since the two regions will cluster separately in de novo clustering.