google-deepmind / alphafold3

AlphaFold 3 inference pipeline.
Other
5.14k stars 578 forks source link

specifying the versions of dependencies #13

Closed anilbey closed 1 week ago

anilbey commented 1 week ago

Hi, could you specify the expected versions (or version ranges) of dependencies like numpy in pyproject.toml file? Or should we install the exact same version provided in requirements.txt and dev-requirements.txt by pip-compile? Thanks

https://github.com/google-deepmind/alphafold3/blob/e2afc41e6028502904aa457892fbcb839436f819/pyproject.toml#L26

charlesbeattie commented 1 week ago

Hi, could you specify the expected versions (or version ranges) of dependencies like numpy in pyproject.toml file?

We specify where we felt it mattered for prediction accuracy.

Or should we install the exact same version provided in requirements.txt and dev-requirements.txt by pip-compile?

Yes I would advise to do that. We provide dev-requirements.txt for reproducibility of the setup we validated. You can follow docker for the install instructions; the relevant part is as follows (with RUNs removed). You do not need requirements.txt as dev-requirements.txt is a super set of requirements.txt that include dependencies needed for the build/install.

# Install the Python dependencies AlphaFold 3 needs.
pip3 install -r dev-requirements.txt
pip3 install --no-deps .
# Build chemical components database (this binary was installed by pip).
build_data