So ANCOM takes a feature table of absolute frequencies, adds a pseudo-count of 1, log transforms the data, and compares the ratio of ASVs between the groups. So, for example, if an ASV has a certain W value, it means that its log-transformed abundance to W other ASVs is significantly different between the groups in total. In reporting the results, is it the absolute abundance of the identified features that are significantly different between the group, or is it its relative abundance?
Since minimum abundance is usually 1.0, and max often several kilos, I suppose it is absolute abundances
I like this quote, from https://www.nature.com/articles/s41467-020-17041-7. Maybe you’ll find it helpful, too, @Parix.
ANCOM which is based on Aitchison’s methodology, uses relative abundances to infer about absolute abundances.
This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.