Sequences that could only be classified at the family level or higher taxonomic ranks should be named Unclassified_Family_1, Unclassified_Family_2, Unclassified_Family_3, etc. so as not to lose taxonomic resolution. Fully agree with this recommendation. Genus is the finest taxonomic resolution rank achieved with 16SrRNA gene sequencing. Therefore, I am unsure whether it is necessary to differentiate each of the sequences that belong to a known bacterial genus. For example, in my taxonomy table, I have 14 sequences belonging to the genus Vibrio. In my results, I find that the genus Vibrio spp. is differentially abundant. This result is due to 7 ASVs belonging to the genus Vibrio, so it would be correct to differentiate these sequences as Vibrio_1, Vibrio_2, Vibrio_3,..... Vibrio_14 to differentiate the specific sequences that influence my results. In the literature, I have seen some analyses where the genus is given and no specific ASV is specified, and in others, only the genus name is given, followed by the specific ASV (e.g., Velionella ASV259).
Hi @DavidRivera,
I think there are three ideas to talk about in htis post.
So, let's start with the idea that taxonomic resolution is the same as sequence identity. Taxonomic identity is a name that we decorate our ASVs with. We assume it represents some kind of biological ontology/hierarchy and that sequences (organisms) we assign similar names to behave in similar ways. (I'm going to stay off my taxonomy soap box right now.) People who kick around the forum a lot know this example already, but I had a study were two ASVs from the same genus (same species based on BLAST!) that had very different behavior and were actually keystone species.
Taxonomic assignment is not an intrensic thing in the ASV or in 16S rRNA sequencing. Our ASVs exist without taxonomic assignments. The resolution I can use to taxonomy is dependent on the database I use, the type of classifier, and the information in the sequence. So, I have enviroment-amplicon-database-classifier combinations where I can accurately and reliably assign species. I have enviroment-amplicon-database-classifier combinations where I can only reliably assign family. It depends on a lot of factors.
That said, I think it's a fair rule of thumb ot say that genus is the finest resolution we can use to decorate ASVs in most circumstances.
I would say this depends on how you decide to test your data! So, if you collapse your data to genus level, you would describe your analysis at genus level. If you analyze your data at ASV level, you should describe your data at ASV level. I would write those differently in my methods and results, so that my reader (and I because I forget) know what level I'm working at.
If I work at genus level, then I'm assuming that taxa with the same name behave the same way regardless of evolutionary relationship. I would report that with something like,
At genus level, we found g. Vibrio was more abundant in condition A vs condition B and g. [Other genus] was less abundant in A vs B (Table SX).
Where Table SX is a supplementary table with all the genera I tested in differentail abundance with the genus level assignment. (See STORMS/STREAMS for details).
If I'm presenting differential abundance based on ASVs, I might describe the data in the text as
There were 7 ASVs mapped to g Vibrio, an ASV from unassigned family 1 and an ASV from genus 2 that were more abundant in condition A vs condition B. We also identifed 8 ASVs from genus 3 which was less abundant in condition A vs condition B (Table SY).
Where again, Table SY gives you the supplemental table with everything
Whether or not you want to identify specific ASVs in the main text depends on the situation and work you're doing. In my example that I linked above, it was really important that people know it was the same ASV that we were finding over and over and over again. There are other cases where it may not matter that people know the specific ASV.
So, long answer to a specific question, with the summary being It DependsTM on what you did and what you want to say.
Best,
Justine
Your lengthy but concise and robust explanation has cleared up my doubts, and I believe that the most accurate way to report my results is at the ASV level. Thank you very much for your time.