I’m not an expert here but I usually run R from Julia itself. The problem with R is that it needs to bring the object to memory and the cube (or even parts of it) are several Gb. So I follow the guides on reading the cube from Julia, extract the layers I’m interested on, do some pre-processing in them already from Julia, and only when I really need statistical tools I do analysis in R with small parts of the dataset.
To do so, I include the following
Pkg.add("RCall") at the top of my Julia file. Once I’m ready to use R, then I call R from Julia with:
<r code here>
You can learn more tricks here: https://github.com/JuliaInterop/RCall.jl