I’m trying to run an analysis that finds associations between bacterial abundance and various behavioural metrics.
I’ve noticed in the scatter plot outputs that my results seem to follow a logistic rather than linear fit, wherein behaviour seems related to either a presence or absence of the bacteria in question (see image for example)
How can I address this in my analysis? Is there any way to make MaAsLin2 fit a logistic rather than linear model? Would this be achieved using a LOGIT transform? Is there any way to force the scatter plots to display and analyze a logistic regression line?
As a side note- I’m using CLR-transformed data, hence the negative values