Hi,
I am performing a 16S analysis and have a question regarding alpha diversity metrics.
When calculating core metrics, you must use the table with all ASVs, for which the phylogenetic tree has been built. This gives alpha diversity values for the samples. Do these values accurately reflect the diversity of the samples? Some publications refer to alpha diversity concerning specific taxonomic levels (genus, family, etc.). Is this the correct approach, or both are ok? Could important information be missed by not calculating alpha diversity at a specific taxonomic level?
Regarding observed features: for example, if you're counting at the genus level, you might have 100 features. However, using a general approach that includes all taxonomic levels might result in a higher count, as one genus could be counted across multiple levels (domain, phylum, class, order, family, genus). Could this potentially overestimate the diversity metrics?
Any suggestions regarding this for a paper publication?
These are all fine approaches. It all comes down to the question of what you are trying to measure. Alpha diversity is measuring the count or diversity of unique observations within a single sample. These observations can be species, genera, families, OTUs, ASVs, genes, functional pathways, metabolites, proteins, ... the list goes on. So you can use alpha diversity metrics to quantify the diversity of observations, but the interpretation must be closely tied to the inputs. In other words, what are you trying to count?
Now this approach is something altogether different. I think it is fine to count the number of unique species or genera or families or whatever. But to combine them together leads to a very blurred result and I would say is totally uninformative and really should not be done. Because these are not actually equivalent units and hence cannot be counted together or compared.
So if you want to calculate alpha divrsity, I recommend that you do it on one metric at a time.