BRACoD: Bayesian Regression Analysis of Compositional Data
Perform convergence tests on the p and beta variables
Install BRACoD in python
Remove NULL values in your OTU and environmental variable
Run the main BRACoD algorithm
Normalize OTU counts and add a pseudo count
Score the results of BRACoD
Simulate microbiome counts
Summarize the results of BRACoD
Threshold your microbiome counts data
The goal of this method is to identify associations between bacteria and an environmental variable in 16S or other compositional data. The environmental variable is any variable which is measure for each microbiome sample, for example, a butyrate measurement paired with every sample in the data. Microbiome data is compositional, meaning that the total abundance of each sample sums to 1, and this introduces severe statistical distortions. This method takes a Bayesian approach to correcting for these statistical distortions, in which the total abundance is treated as an unknown variable. This package runs the python implementation using reticulate.