PaccMann/paccmann_gp

PyTorch/skopt based implementation of Bayesian optimization with Gaussian processes - build to optimize latent spaces of VAEs to generate molecules with desired properties

paccmann_gp

Bayesian Optimisation with Gaussian Processes for molecular generative models.

Installation

Create a conda environment:

conda env create -f conda.yml

Activate the environment:

conda activate paccmann_gp

Example usage

In the examples directory is an example script example.py that makes use of paccmann_gp for a combined optimisation for QED, SAscore and affinity to the transcription factor ERG.

(paccmann_gp) $ python examples/example.py -h
usage: example.py [-h]
                    svae_path affinity_path
                    optimisation_name


positional arguments:
  svae_path          Path to downloaded SVAE model.
  affinity_path      Path to the downloaded affinity prediction model.
  optimisation_name  Name for the optimisation.

The trained SVAE and affinity models can be downloaded from the SELFIESVAE and affinity folders located here.

Citation

If you use this repo in your projects, please temporarily cite the following:

@article{born2022active,
	author = {Born, Jannis and Huynh, Tien and Stroobants, Astrid and Cornell, Wendy D. and Manica, Matteo},
	title = {Active Site Sequence Representations of Human Kinases Outperform Full Sequence Representations for Affinity Prediction and Inhibitor Generation: 3D Effects in a 1D Model},
	journal = {Journal of Chemical Information and Modeling},
	volume = {62},
	number = {2},
	pages = {240-257},
	year = {2022},
	doi = {10.1021/acs.jcim.1c00889},
	note ={PMID: 34905358},
	URL = {https://doi.org/10.1021/acs.jcim.1c00889}
}