Currently, I am struggling with the right strategy for performing analyses related to the functional annotations (KEGGs).
Among the KEGGs I am getting, some are not related to the bacteria (code “E.AniM”, for example). My assumption that all such KEGGs are from the host or feed, rather than microbial communities, and should be filtered out before the analyses as contaminants.
Am I missing something by doing it like that? Is there any reason to keep them in the dataset? Or my interpreation of the code (E.AniM) is incorrect?
Correct, it looks like the code E.AniM is for mammals specifically. So I agree, these are probably host or feed or environmental contaminants. Depending on your experimental goals, these might be interesting (e.g., if looking at dietary components), or could probably be safely ignored.
Although I was initially inclined to filter out such functions from the dataset, I decided to ask this question here to raise the issue of host/feed DNA residuals in the cleaned reads. Depending on the host and sample type, for some datasets, it is problematic to delete all the DNA that is actually the host's or the feed's DNA. This may lead to the detection of functions that are not of microbial origin. Assuming the user is not interested in them, shouldn’t there be an optional step to remove such functions from the egg-nog output before converting it into other categories (pathways, reactions, etc.) to avoid associated changes in finall counts? Or did I miss it, and it is already implemented?
with the right strategy for performing analyses related to the functional annotations (KEGGs).