Accelerate your chemistry with ReactWise

ORDerly: Datasets and benchmarks for chemical reaction data

ORDerly is an open-source Python package for customizable and reproducible preparation of reaction data stored in accordance with the increasingly popular Open Reaction Database (ORD) schema. We use ORDerly to clean US patent data stored in ORD and generate datasets for forward prediction, retrosynthesis, as well as the first benchmark for reaction condition prediction. By providing a customizable open-source solution for cleaning and preparing large chemical reaction data, ORDerly is poised to push forward the boundaries of machine learning applications in chemistry.

Accelerated chemical reaction optimization using multi-task learning

Functionalization of C–H bonds is a key challenge in medicinal chemistry, particularly for fragment-based drug discovery.  We explore the use of multitask Bayesian optimization (MTBO) for real-world medicinal chemistry applications using an autonomous flow-based reactor platform. The use of the MTBO algorithm demonstrated an efficient optimization strategy with large potential cost reductions when compared to industry-standard process optimization techniques, representing a step-change in the utilization of data and machine learning with the goal of accelerated reaction optimization.

A Brief Introduction to Chemical Reaction Optimization

Identification of optimal chemical reaction parameters involving continuous (time, temperature, pressure) and categorical variables (catalyst, solvent, reagent) is a non-trivial challenge. This review highlights the basics, and the cutting-edge chemical reaction optimization - from one-factor-at-a-time (OFAT) over design of experiments (Doe) to Bayesian optimization. The review details the relation of reaction optimization to process scale-up and can thereby serve as a reference for inspired scientists for each of these techniques, detailing several of their respective applications.

The effect of chemical representation on active machine learning towards closed-loop optimization

Multivariate chemical reaction optimization involving catalytic systems is a non-trivial task due to the high number of continuous parameters and discrete choices. The translation of chemical reaction conditions into a machine-readable format requires the identification of highly informative features which accurately capture the factors which determine reaction success. Herein, we compare the efficacy of different calculated chemical descriptors and demonstrated that in small datasets generic featurization can outperform complex descriptors within a closed-loop reaction optimization.

Automated ph adjustment driven by robotic workflows and active machine learning

Buffer solutions have tremendous importance in biological systems and in formulated products. Modelling the pH response for multi-buffered poly-protic systems is a challenge. Combining ML-driven closed-loop optimization and robotic workflows, we present an autonomous strategy for small scale batch pH adjustment. Integration of prior data into the optimization protocol proved to be a successful strategy in making the process even more efficient, offering a versatile and efficient strategy for a pH adjustment processes.

Quantitative In Silico Prediction of the Rate of Protodeboronation by a Mechanistic Density Functional Theory-Aided Algorithm

The Suzuki reaction represents one of the most versatile and relevant reactions towards drug manufacturing. However, understanding how to avoid the protodeboronation side reaction remains a challenge. We developed an algorithm for in silico prediction of the rate of proto-deboronation. Using quantum mechanical calculation a general mechanistic model is presented with the potential to provide great assistance to chemists performing reactions that feature boronic acids within an academic and industrial setting.