scVI - environment management


I was wondering what environment you recommend using in order to keep analysis reproducible over time. I was having some issues with pipenv and recently switched over to pip-tools. Do you have any recommendations?

Thank you!


I personally use pyenv and pyenv virtualenv, I had some trouble with pipenv in the past. That said, conda (miniconda) is probably the most popular option, as it’s also the easiest way to mix R and python dependencies.

Thank you for the quick response!
Do you have any advice for virtualenv management in this situation: you start with one set of packages and record their versions (scvi-tools==0.8.1). Then you need the functionality that got added later, and you install scvi-tools==0.14.0, for example. How to manage updates like that and updates of dependencies?

I think you can use the no dependencies flag for pip install, and then see which packages/versions are actually in need of upgrade.