🌻 BOUQUET 💐

Generate beautiful ensembles of molecular geometries Bayesian Optimization Using QUantum Energy Tool

This repo contains code for optimizing conformers using Bayesian optimization for active learning and quantum chemistry computations.

Background

Conformers define the different geometries with the same molecular bonding graph but different coordinates. Finding the lowest-energy conformation is a common task in molecular modeling, and one that often requires significant time to solve. We implement optimal experimental design techniques to solve this problem following recent work that uses Bayesian optimization find optimize dihedral angles.

Installation

Build the environment using anaconda:

conda env create --file environment.yml --force

Or use pixi:

pixi install

The key parts of the code are available through pip:

pip install '.[all]'

Use

bouquet provides a simple interface to the code. To optimize cysteine with default arguments.

bouquet --smiles "C([C@@H](C(=O)O)N)S"

This will produce a folder in the solutions directory containing the optimized geometry (final.xyz) and many other files for debugging.

Call bouquet --help for available options.