/edbo

Experimental Design via Bayesian Optimization

Primary LanguageJupyter NotebookMIT LicenseMIT

edbo

Experimental Design via Bayesian Optimization: edbo is a practical implementation of Bayesian optimization for chemical synthesis.

Reference: Shields, Benjamin J.; Stevens, Jason; Li, Jun; Parasram, Marvin; Damani, Farhan, Janey, Jacob; Adams, Ryan P.; Doyle, Abigail G. "Bayesian Reaction Optimization as A Tool for Chemical Synthesis" Manuscript Accepted.

Documentation: https://b-shields.github.io/edbo/index.html

Installation

(0) Create anaconda environment

conda create --name edbo python=3.7.5

(1) Install rdkit, Mordred, and PyTorch

conda activate edbo
conda install -c rdkit rdkit
conda install -c rdkit -c mordred-descriptor mordred
conda install -c pytorch pytorch=1.3.1

(2) Install EDBO

pip install edbo

Running Notebooks

conda install jupyterlab