Firstly, I have looking in the forum (in case someone had had my same doubt), but I didn't find it. If this has been commented in the past, please, could you indicate the posts? Thanks in advance!
But, in case it has not been brought up before:
I am a bit confuse about the filtering processing after importing phyloseq object. I know that if you want to remove contaminants (for example) of your phyloseq object, you should find them and then, removing them from the original table by OTU ID instead of by the taxonomy target level you were checking (for example, family level).
However, I would like to understand why is that. And read more about it. Could you recommend me some reliable sources or articles?
Filtering contamination is controversial because there are different sources of unwanted 'contamination' and it's nearly impossible to separate the sources of observed variation and get rid of just the 'bad' ones you don't want.
Check out this discussion of contamination and cross-talk on this best-of-the-forums thread: