Absence as a peculiarity


Sometimes when some group of samples is significantly different from others it is not due to the presence of some specific OTU’s, but because of absence of some OTU’s that are very common in all other groups, but absent in this.

Do we have any analysis that highlight those?
Like lefse highlight otu’s that are specific for given group, but the one that would highlight those the absence of which is specific?
Or the one that shows both tipes of peculiarities. With + and -.

Thank you.

We use things like decision trees, logistic regression, and random forest for this type of situation, which is pretty common in microbial ecology datasets. We outline these approaches in a recent mbio paper:


1 Like

Ok. Thank you.

For calculating which of them do you have tools in mothur?

It’s done with mothur output (i.e. shared files) fed into R