I understand and agree with the idea that unclassified ASVs are also informative and their elimination wastes valuable information. However, these sequences without a clear taxonomic assignment (i.e. from family to lower taxonomic ranks ) would not disrupt the phylogeny of a particular group of samples and therefore the alpha and beta diversity metrics that incorporate phylogenetic distance as a parameter for determining diversity (i.e. Faiths PD and Unifrac distances).
Hi @DavidRivera,
Welcome to the 
forum!
I'm assuming that you're assigning taxonomy through a naive bayesian classifier and either building a de novo tree or doing fragment insertion. If this is the case, then your tree and taxonomy don't need to agree. I can assign my taxonomy with RDP and do a fragment insertion into the gg13_8 backbone. I can use Silva for my taxonomy and do a de novo tree. I could do fragment insertion into Silva 128 and decide that I dont care about taxonomy today because all I really need right now is a phylogenetic tree. You tree will be based on the sequences you provide, regardless of how you decide to name them. So, whether you got a genus level assignment or not, the sequence will be the same, and you can move forward with your phylogenetic diversity metrics.
Best,
Justine
4 Likes
Thanks for your quick and accurate response.
1 Like