Predictive chemistry: machine learning for reaction deployment, reaction development, and reaction discovery

Tu, Zhengkai; Stuyver, Thijs; Coley, Connor W.

doi:10.1039/d2sc05089g

Cited by 60 publications

(45 citation statements)

References 204 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Notably, our retriever is a novel rule-based module developed to retrieve reactant molecules directly from a reaction database using product molecules as queries. Existing retrieval-based models retrieve reactant molecules from a predefined molecule set using deep-learning-based methods [39][40][41] . READRetro consists of a rule-based retriever and a deep-learning-based model, so it can fully leverage both the memorability of the retriever and generalizability of the deep learning model.…”

Section: Evaluation Of the Multi-step Retrosynthesismentioning

confidence: 99%

READRetro: Natural Product Biosynthesis Planning with Retrieval-Augmented Dual-View Retrosynthesis

Lee

Kim

Choi

et al. 2023

Preprint

View full text Add to dashboard Cite

Elucidating the biosynthetic pathways of natural products has been a major focus of biochemistry and pharmacy. However, predicting the whole pathways from target molecules to metabolic building blocks remains a challenge. Here we propose READRetro as a practical bio-retrosynthesis tool for planning the biosynthetic pathways of natural products. READRetro effectively resolves the tradeoff between generalizability and memorability in bio-retrosynthesis by implementing two separate modules; each module is responsible for either generalizability or memorability. Specifically, READRetro utilizes a rule-based retriever for memorability and an ensemble of two dual-representation-based deep learning models for generalizability. Through extensive experiments, READRetro was demonstrated to outperform existing models by a large margin in terms of both generalizability and memorability. READRetro was also capable of predicting the known pathways of complex plant secondary metabolites such as monoterpene indole alkaloids, demonstrating its applicability in the real-world bio-retrosynthesis planning of natural products. A website (https://readretro.net) and open-source code have been provided for READRetro, a practical tool with state-of-the-art performance for natural product biosynthesis research.

show abstract

Section: Evaluation Of the Multi-step Retrosynthesismentioning

confidence: 99%

READRetro: Natural Product Biosynthesis Planning with Retrieval-Augmented Dual-View Retrosynthesis

Lee

Kim

Choi

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…The automated generation of reaction arrays to optimize or discover a coupling between two substrates is a contemporary problem. [6][7][8][9][10][11][12] Recently, generative transformers, a form of artificial intelligence (AI), have emerged as interactive language models that can interpret and answer scientific questions via verbal human input. 13,14 Herein we demonstrate how the general-purpose language model ChatGPT can be utilized to generate initialguess reaction array designs for specific substrate pairs.…”

Section: Introductionmentioning

confidence: 99%

Designing Chemical Reaction Arrays using phactor and ChatGPT

Mahjour

Hoffstadt

Cernak

2023

Preprint

View full text Add to dashboard Cite

High throughput experimentation (HTE) is a common practice in the pharmaceutical industry. Medicinal chemists design reaction arrays to optimize the yield of couplings between building blocks and/or pharmacophores. Popular reactions attempted by medicinal chemists include the amide coupling, Suzuki coupling, and Buchwald-Hartwig coupling. We show how the artificial intelligence (AI) language model ChatGPT can automatically formulate reaction arrays for these common reactions based on the literature corpus it was trained on. Furthermore, we showcase how ChatGPT results can be directly translated into inputs for the HTE management software phactor, which enables automated execution and analysis of assays. This workflow is experimentally demonstrated.

show abstract

“…With the advent of machine learning techniques and the digitization of much of the chemical literature and patents, new approaches to retrosynthetic analyses are being developed. For a recent review see Reference [4 ]. Most of the retrosynthetic analysis tools are based on experimental data.…”

Section: Introductionmentioning

confidence: 99%

Accelerating Reaction Network Explorations with Automated Reaction Template Extraction and Application

Unsleber

2023

Preprint

View full text Add to dashboard Cite

The autonomous exploration of chemical reaction networks with first-principles methods generates vast amounts of data. Especially explorations that explore reaction networks autonomously and without tight constraints run the risk of exploring regions of reaction space that are not of interest. Consequently, the required human time for analysis and computer time for data generation can make these explorations unfeasible. Here, we show how an automated extraction of reaction templates can facilitate the transfer of chemical knowledge from existing data. This process significantly accelerates reaction network explorations and improves cost-effectiveness. The reaction templates allow for a simple steering mechanism in autonomous reaction network explorations, which we exemplify with a polymerization reaction. We discuss definitions of reaction templates and their generation based on molecular graphs. Graph matching and sub-graph searches based on molecular graphs and reaction templates may allow data clustering and new analyses of the generated reaction network.

show abstract

Predictive chemistry: machine learning for reaction deployment, reaction development, and reaction discovery

Abstract: The field of predictive chemistry relates to the development of models able to describe how molecules interact and react. It encompasses the long-standing task of computer-aided retrosynthesis, but is far...

Cited by 60 publications

References 204 publications

READRetro: Natural Product Biosynthesis Planning with Retrieval-Augmented Dual-View Retrosynthesis

READRetro: Natural Product Biosynthesis Planning with Retrieval-Augmented Dual-View Retrosynthesis

Designing Chemical Reaction Arrays using phactor and ChatGPT

Accelerating Reaction Network Explorations with Automated Reaction Template Extraction and Application

Contact Info

Product

Resources

About