Is there a way to use RDKit without Anaconda or Conda?

Question

RDKit is a Python library for chemistry. I don't want to use an Anaconda distribution, and I am not sure (from the documentation: https://www.rdkit.org/docs/Install.html ) if there is a way to use the package without it. Can someone give me some pointers on installing and using this without conda or installing an Anaconda distribution?

Answer 1

The documentation does provide directions for installing from alternative repositories (e.g, homebrew) or building from source.

Nevertheless, Conda is the simplest path, and just because you don't want Anaconda (neither do I!), that shouldn't deter you from enjoying the benefits of deep dependency management that Conda provides. There are minimal and performant distributions, like Mambaforge, that users only focused on environment and package management often find preferable. Also, don’t use the base environment for work, but create a new one, like

mamba create -n rdkit rdkit

Answer 2

I just installed rdkit and then installed jupyter (without anaconda), and it is working. It works in PyCharm too, but I'm not able to display images of molecules as SVG files.

Answer 3

I'm also not a fan of conda as it is a half-step to installing software in docker images, which is a better solution. And the shell heroics, while clever, are clunky. And installs are crazy slow.

You should be able to do a direct install w/ pip:

python3 -m pip install rdkit

This will install it into wherever the current python3 interpreter has its default modules, without conda. You may need to run with sudo. You can install it into a virtualenv with:

python3 -m venv env
. env/bin/activate
python3 -m pip install rdkit

Note that the examples above work with any recent version of python3, and circumvent the pip CLI which often leaves it ambiguous which python3 version the install is applied-to.