Given the developments and increased access to machine learning algorithms such as LDA that have happened since LEfSe was developed, is there now a way to perform the same analysis with standard tools?
The LEfSe code base is in pretty primitive python 2.7 and is not even organized as a package--just a collection of scripts. You can use the Galaxy interface provided by the Huttenhower lab, but I seem to run into problems when using covariates.
I cloned LEfSe from bitbucket and tried running the example script and threw errors. The repository doesn't have a place to raise issues. I plan to email the author.
Is there a simple way to do the same analysis that LEfSe does with stock packages, but without having to delve into the original paper and re-implement on my own?
