Graph neural networks for materials science and chemistry

Communications Materials, Journal Year: 2022, Volume and Issue: 3(1)

Published: Nov. 26, 2022

Abstract Machine learning plays an increasingly important role in many areas of chemistry and materials science, being used to predict properties, accelerate simulations, design new structures, synthesis routes materials. Graph neural networks (GNNs) are one the fastest growing classes machine models. They particular relevance for as they directly work on a graph or structural representation molecules therefore have full access all relevant information required characterize In this Review, we provide overview basic principles GNNs, widely datasets, state-of-the-art architectures, followed by discussion wide range recent applications GNNs concluding with road-map further development application GNNs.

Language: Английский

---

Structure-based drug design with geometric deep learning

Current Opinion in Structural Biology, Journal Year: 2023, Volume and Issue: 79, P. 102548 - 102548

Published: Feb. 25, 2023

Structure-based drug design uses three-dimensional geometric information of macromolecules, such as proteins or nucleic acids, to identify suitable ligands. Geometric deep learning, an emerging concept neural-network-based machine has been applied macromolecular structures. This review provides overview the recent applications learning in bioorganic and medicinal chemistry, highlighting its potential for structure-based discovery design. Emphasis is placed on molecular property prediction, ligand binding site pose de novo The current challenges opportunities are highlighted, a forecast future presented.

Language: Английский

---

Past, Present, and Future Perspectives on Computer-Aided Drug Design Methodologies

Molecules, Journal Year: 2023, Volume and Issue: 28(9), P. 3906 - 3906

Published: May 5, 2023

The application of computational approaches in drug discovery has been consolidated the last decades. These families techniques are usually grouped under common name "computer-aided design" (CADD), and they now constitute one pillars pharmaceutical pipelines many academic industrial environments. Their implementation demonstrated to tremendously improve speed early steps, allowing for proficient rational choice proper compounds a desired therapeutic need among extreme vastness drug-like chemical space. Moreover, CADD allows rationalization biochemical interactive processes interest at molecular level. Because this, tools extensively used also field 3D design optimization entities starting from structural information targets, which can be experimentally resolved or obtained with other computer-based techniques. In this work, we revised state-of-the-art computer-aided methods, focusing on their different scenarios biological interest, not only highlighting great potential benefits, but discussing actual limitations eventual weaknesses. This work considered brief overview methods discovery.

Language: Английский

---

Enabling late-stage drug diversification by high-throughput experimentation with geometric deep learning

Nature Chemistry, Journal Year: 2023, Volume and Issue: 16(2), P. 239 - 248

Published: Nov. 23, 2023

Abstract Late-stage functionalization is an economical approach to optimize the properties of drug candidates. However, chemical complexity molecules often makes late-stage diversification challenging. To address this problem, a platform based on geometric deep learning and high-throughput reaction screening was developed. Considering borylation as critical step in functionalization, computational model predicted yields for diverse conditions with mean absolute error margin 4–5%, while reactivity novel reactions known unknown substrates classified balanced accuracy 92% 67%, respectively. The regioselectivity major products accurately captured classifier F -score 67%. When applied 23 commercial molecules, successfully identified numerous opportunities structural diversification. influence steric electronic information performance quantified, comprehensive simple user-friendly format introduced that proved be key enabler seamlessly integrating experimentation functionalization.

Language: Английский

---

Prospective de novo drug design with deep interactome learning

Nature Communications, Journal Year: 2024, Volume and Issue: 15(1)

Published: April 22, 2024

Abstract De novo drug design aims to generate molecules from scratch that possess specific chemical and pharmacological properties. We present a computational approach utilizing interactome-based deep learning for ligand- structure-based generation of drug-like molecules. This method capitalizes on the unique strengths both graph neural networks language models, offering an alternative need application-specific reinforcement, transfer, or few-shot learning. It enables “zero-shot" construction compound libraries tailored bioactivity, synthesizability, structural novelty. In order proactively evaluate interactome framework protein design, potential new ligands targeting binding site human peroxisome proliferator-activated receptor (PPAR) subtype gamma are generated. The top-ranking designs chemically synthesized computationally, biophysically, biochemically characterized. Potent PPAR partial agonists identified, demonstrating favorable activity desired selectivity profiles nuclear receptors off-target interactions. Crystal structure determination ligand-receptor complex confirms anticipated mode. successful outcome positively advocates de application in bioorganic medicinal chemistry, enabling creation innovative bioactive

Language: Английский

---

Informing geometric deep learning with electronic interactions to accelerate quantum chemistry

Proceedings of the National Academy of Sciences, Journal Year: 2022, Volume and Issue: 119(31)

Published: July 28, 2022

Predicting electronic energies, densities, and related chemical properties can facilitate the discovery of novel catalysts, medicines, battery materials. However, existing machine learning techniques are challenged by scarcity training data when exploring unknown spaces. We overcome this barrier systematically incorporating knowledge molecular structure into deep learning. By developing a physics-inspired equivariant neural network, we introduce method to learn representations based on interactions among atomic orbitals. Our method, OrbNet-Equi, leverages efficient tight-binding simulations learned mappings recover high-fidelity physical quantities. OrbNet-Equi accurately models wide spectrum target while being several orders magnitude faster than density functional theory. Despite only using samples collected from readily available small-molecule libraries, outperforms traditional semiempirical learning-based methods comprehensive downstream benchmarks that encompass diverse main-group processes. also describes in challenging charge-transfer complexes open-shell systems. anticipate strategy presented here will help expand opportunities for studies chemistry materials science, where acquisition experimental or reference is costly.

Language: Английский

---

End-to-end differentiable construction of molecular mechanics force fields

Chemical Science, Journal Year: 2022, Volume and Issue: 13(41), P. 12016 - 12033

Published: Jan. 1, 2022

Graph neural network-based continuous embedding is used to replace a human expert-derived discrete atom typing scheme parametrize accurate and extensible molecular mechanics force fields.

Language: Английский

---

Machine Learning for Fast, Quantum Mechanics-Based Approximation of Drug Lipophilicity

ACS Omega, Journal Year: 2023, Volume and Issue: 8(2), P. 2046 - 2056

Published: Jan. 4, 2023

Lipophilicity, as measured by the partition coefficient between octanol and water (log P), is a key parameter in early drug discovery research. However, measuring log P experimentally difficult for specific compounds ranges. The resulting lack of reliable experimental data impedes development accurate silico models such compounds. In certain projects at Novartis focused on compounds, quantum mechanics (QM)-based tool estimation has emerged valuable supplement to measurements preferred alternative existing empirical models. this QM-based approach incurs substantial computational cost, limiting its applicability small series prohibiting quick, interactive ideation. This work explores set machine learning (Random Forest, Lasso, XGBoost, Chemprop, Chemprop3D) learn calculated values both public an in-house obtain computationally affordable, lipophilicity. message-passing neural network model Chemprop best performing with mean absolute errors 0.44 0.34 units scaffold split test sets sets, respectively. Analysis curves suggests that further decrease error can be achieved increasing training size. While directly trained perform better approximating determined than values, we discuss potential advantages using going beyond limits quantitation. We analyze impact splitting strategy gain insights into failure modes. Potential use cases presented include pre-screening large compound collections prioritization full QM calculations.

Language: Английский

---

From intuition to AI: evolution of small molecule representations in drug discovery

Briefings in Bioinformatics, Journal Year: 2023, Volume and Issue: 25(1)

Published: Nov. 22, 2023

Abstract Within drug discovery, the goal of AI scientists and cheminformaticians is to help identify molecular starting points that will develop into safe efficacious drugs while reducing costs, time failure rates. To achieve this goal, it crucial represent molecules in a digital format makes them machine-readable facilitates accurate prediction properties drive decision-making. Over years, representations have evolved from intuitive human-readable formats bespoke numerical descriptors fingerprints, now learned capture patterns salient features across vast chemical spaces. Among these, sequence-based graph-based small become highly popular. However, each approach has strengths weaknesses dimensions such as generality, computational cost, inversibility for generative applications interpretability, which can be critical informing practitioners’ decisions. As discovery landscape evolves, opportunities innovation continue emerge. These include creation high-value, low-data regimes, distillation broader biological knowledge novel modeling up-and-coming therapeutic modalities.

Language: Английский

---

In Silico Chemical Experiments in the Age of AI: From Quantum Chemistry to Machine Learning and Back

Advanced Materials, Journal Year: 2024, Volume and Issue: 36(30)

Published: May 25, 2024

Abstract Computational chemistry is an indispensable tool for understanding molecules and predicting chemical properties. However, traditional computational methods face significant challenges due to the difficulty of solving Schrödinger equations increasing cost with size molecular system. In response, there has been a surge interest in leveraging artificial intelligence (AI) machine learning (ML) techniques silico experiments. Integrating AI ML into increases scalability speed exploration space. remain, particularly regarding reproducibility transferability models. This review highlights evolution from, complementing, or replacing energy property predictions. Starting from models trained entirely on numerical data, journey set forth toward ideal model incorporating physical laws quantum mechanics. paper also reviews existing their intertwining, outlines roadmap future research, identifies areas improvement innovation. Ultimately, goal develop architectures capable accurate transferable solutions equation, thereby revolutionizing experiments within materials science.

Language: Английский

---