Introduction

This repository includes an open-source implementation of the Site of Borylation (SoBo) model that predicts the regioselectivity for the iridium-catalyzed borylation. The manuscript is available on ChemRxiv and JACS.

Environment setup

Start with setting up a new conda virtual environment (venv) (see here how to setup Miniconda)

> conda create --name sobo python=3.9

and activate the created venv

> conda activate sobo

Install dependencies

Get openbabel (3.1.0) dependency

> conda install -c conda-forge openbabel

Get xtb dependency (6.4.0) by downloading a precompiled binary from GitHub, by compiling the source code yourself, or by a conda installation (only fixed versions available)

> conda install -c conda-forge xtb

Then install the SoBo (0.2.0) method via pip

> pip install sobo

The typical installation time on a standard desktop computer is in the range of minutes. Several examples are available with expected output inside the examples directory.

Run predictions

Once the model is setup (check reference outputs in examples) you can perform predictions by entering a SMILES string and a name for the system of interest.

> sobo --smiles <SMILES> --name <NAME>

You can get additional information via

> sobo --help

Acknowledgements

Johan Westin restructured the package structure of the SoBo method, which has been adapted for the purpose of creating a PyPi package.

Reference

If you want to apply or reference this work please always cite:

Caldeweyher, Eike and Elkin, Masha and Gheibi, Golsa and Johansson, Magnus and Sköld, Christian and Norrby, Per-Ola and Hartwig, John F., J. Am. Chem. Soc, 2023. DOI: 10.1021/jacs.3c04986

@article{doi:10.1021/jacs.3c04986,
author = {Caldeweyher, E. and Elkin, M. and Gheibi, G. and Johansson, M. J. and Sköld, C. and Norrby, P-O and Hartwig, J. F.},
title = {Hybrid Machine Learning Approach to Predict the Site Selectivity of Iridium-Catalyzed Arene Borylation},
journal = {J. Am. Chem. Soc.},
year = {2023},
doi = {10.1021/jacs.3c04986},
}