OTU vs ESV

[lukazuvic]

Hi Till,
I have a question. I am working on DNA metabrcoding of stomach contents of Atlantic bluefin tuna and also marine sediments collected beneath Atlantic bluefin tuna cages, where the organisms I am interested in are metazoans. I was wondering if you would suggest OTU clustering or ESV denoising in this case, and if you recommend OTU clustering, can I use an older version of Apscale in which OTU clustering is still available?

Thank you very much and best regards,
Luka

[TillMacher]

Hi Luka,

that depends mostly on your research question and on the reference database situation. If you work on species-level you can use either ESVs or OTUs, since the analysis-level will be the taxonomic assignment anyways. If you expect a low overall taxonomic resolution in general but still want to work with species approximations, then OTUs might be more suitable. ESVs might be interesting when looking for intraspecific patters.

However, I know from my data that OTUs often do not correspond that well with species (i.e., one species can have multiple OTUs). Furthermore, ESVs (as they are generated in Apscale) have the advantage that they are time-stable and can be compared to future datasets without having the need to run the raw data processing again. OTUs and their distribution on the other hand will change when adding or removing samples.

Overall, I would recommend generating ESVs, assign taxonomy, and try to work with the taxonomy. If the resolution is too low, then work with ESVs (which are kind of haplotypes) but keep in mind that this is rather a proxy for genetic diversity and not species diversity. Also, ESV richness can be inflated by a couple of species with high intraspecific distances in your marker region. But so can also be OTUs.

I guess there is no definite answer to your question, but I hope that helps anyways! :)

cheers
Till

[lukazuvic]

Hi Till,
Thank you for your answer. In case of metabarcoding of stomach contents, I want to detect prey species and compare them with morphological approach findings to better understand the feeding ecology of tuna, while in case of marine sediments I want to detect species that indicate environmental impacts of aquaculture. I am working with COI and 18S region, and this is the taxonomic resolution and richness for COI amplified stomach contents dataset where I used ESV denoising and that was annotated with Midori database. And if looking these charts, there are many ESVs that belong to the same species. So maybe I can try OTU clustering with older version of Apscale and see what will I get and then compare results.

BR,
Luka