Im not sure where the best place is to put this, but I figure this is as good as anywhere, in the hopes that someone else has been more successful that I have, or that I can save someone else from two days with an rapidly increasing FMP rate.
In general, I try to put all my analysis information for a project into a single conda environment and then do my analysis in Jupyter Notebooks. That way, Ive got a collection of documented scripts that can, in theory, re-constituted easily.
Ive been trying to get an R kernel for an environment built with QIIME2-2018. After two days, Ive isolated the issue to a mixture of vanilla R (which I seem to need for the Jupyter kernel) and conda-forge. Its not recommended to merge the two environments.
So, I guess my question here is whether anyone has had success in building an environment with a recent qiime2 build, and an R kernel. And, if so, Id love it if you could share your recipe for avoiding the conda forge/conda default trap.