The use of machine learning methods for the prediction of reaction yield is an emerging area. We demonstrate the applicability of support vector regression (SVR) for predicting reaction yields, using combinatorial data. Molecular descriptors used in regression tasks related to chemical reactivity have often been based on time-consuming, computationally demanding quantum chemical calculations, usually density functional theory. Structure-based descriptors (molecular fingerprints and molecular graphs) are quicker and easier to calculate and are applicable to any molecule. In this study, SVR models built on structurebased descriptors were compared to models built on quantum chemical descriptors. The models were evaluated along the dimension of each reaction component in a set of Buchwald-Hartwig amination reactions. The structure-based SVR models outperformed the quantum chemical SVR models, along the dimension of each reaction component. The applicability of the models was assessed with respect to similarity to training. Prospective predictions of unseen Buchwald-Hartwig reactions are presented for synthetic assessment, to validate the generalizability of the models, with particular interest along the aryl halide dimension.
The concepts behind targeting waters for potency and selectivity gains have been well documented and explored, although maximizing such potential gains can prove to be challenging. This problem is exacerbated in cases where there are multiple interacting waters, wherein perturbation of one water can affect the free energy landscape of the remaining waters. Knowing the right modification a priori is challenging, and computational approaches are ideally suited to help answer the key question of which substitution is best to try. Here, we use Grand Canonical Monte Carlo and the recent Grand Canonical Alchemical Perturbation methods to both understand and predict the effect of ligand-mediated water displacement when more than one water molecule is involved, as well as to understand how exploiting water networks can help govern selectivity.
The use of machine learning methods for the prediction of reaction yield is an emerging area. We demonstrate the applicability of support vector regression (SVR) for predicting reaction yields, using combinatorial data. Molecular descriptors used in regression tasks related to chemical reac?tivity have often been based on time-consuming, computationally demanding quantum chemical calculations, usually density functional theory. Structure-based descriptors (molecular fingerprints and molecular graphs) are quicker and easier to calculate, and are applicable to any molecule. In this study, SVR models built on structure-based descriptors were compared to models built on quantum chemical descriptors. The models were evaluated along the dimension of each reaction component in a set of Buchwald-Hartwig amination reactions. The structure-based SVR models out-performed the quantum chemical SVR models, along the dimension of each reaction compo?nent. The applicability of the models was assessed with respect to similarity to training. Prospec?tive predictions of unseen Buchwald-Hartwig reactions are presented for synthetic assessment, to validate the generalisability of the models, with particular interest along the aryl halide dimension.
The synthesis of new molecules is essential for progress in various sectors within the chemical industry and academia. Medicinal and materials chemistry are two examples. Searching through vast regions of chemical space for routes to new molecules is a time-consuming process carried out by expert synthetic chemists. The use of machine learning and artificial intelligence for synthetic chemistry is rapidly expanding, the aim being to reduce the timelines of chemical syntheses. Tools, which predict products of chemical reactions and design retrosynthetic routes, are attracting particular attention. Emerging computer-aided synthesis design (CASD) programs are not intended to replace synthetic chemists but to aid them in everyday decision making. The incorporation of condition optimisation and reaction performance is highly desirable. Combining such tools with an automated synthesis testing module holds much promise for the future of reaction condition optimisation. To achieve the desired progress in, and acceptance of CASD, there are a few challenges that need to be addressed.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.