I have 16s paired end data that I'm trying to analyse but I've been struggling to define/understand the output from the vsearch command. This is my first time programming/using bioinformatics and my questions might be silly.
I've understood that you can either work with ASVs or OTUs so I tried both to see how they differ. To generate my FeatureTables and RepSeqs i did the following for the two:
ASV analysis - DADA2
OTU analysis - vsearch (de novo, 97%) with DADA2 input
Now my question is, if using OTU clustering with DADA2 input, doesn't that generate a FeatureTable and RepSeqs that are some type of cross between OTUs and ASVs. To my understanding, with the vsearch command I'm clustering ASVs, which would mean I don't get a classic OTU table? When using this method, to I refer to it being a product from ASVs or OTUs?
The diversity measurments varied between the ASV and OTU with DADA2 input and I wonder if it would vary if I made a third beta diversity with a classic OTU table?
Both OTUs and ASVs are forms of feature tables. DADA2 is a denoising technique and works to remove error and return to the original sequence. This gives you single nucleotide resolution along the lenght of your. sequence. OTU clustering can then take those tidied sequences and cluster them; in this case, DADA2 serves as an enhanced quality filtering technique.
There's a really nice thread from a while back where a lot of us discussed our opinions on denoising, clustering, and what's the best approach. In general, different people different views, (my summary is that most mods are pro-denoising but it is worth a read). It may also be worthwhile reading a benchmarking paper, I think Denoising the denoisers is fairly comprehensive.
