Inverse molecular design using machine learning: Generative models for matter engineering

By Benjamín Sánchez-Lengeling, Alán Aspuru-Guzik. 2018

1,800 citations recorded by the source corpus.

The discovery of new materials can bring enormous societal and technological progress. In this context, exploring completely the large space of potential materials is computationally intractable. Here, we review methods for achieving inverse design, which aims to discover tailored materials from the starting point of a particular desired functionality. Recent advances from the rapidly growing field of artificial intelligence, mostly from the subfield of machine learning, have resulted in a fertile exchange of ideas, where approaches to inverse molecular design are being proposed and employed at a rapid pace. Among these, deep generative models have been applied to numerous classes of materials: rational design of prospective drugs, synthetic routes to organic compounds, and optimization of photovoltaics and redox flow batteries, as well as a variety of other solid-state materials.

Subfields: Computational Theory and Mathematics, Environmental Chemistry, Materials Chemistry, Generative Models, Computational Chemistry & Drug Discovery

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Theory and Mathematics

In Environmental Chemistry

In Materials Chemistry

In Generative Models

In Computational Chemistry & Drug Discovery

Citations in the corpus

Emerging materials intelligence ecosystems propelled by machine learning (2020, methodology): “We pay special attention to the protocols for data acquisition and management, context-dependent representation of data, transformation of data to surrogate predictive models and knowledge, and emerging strategies for solving inverse problems 6 that can autonomously drive a materials laboratory.”
Integrating machine learning and multiscale modeling—perspectives, challenges, and opportunities in the biological, biomedical, and behavioral sciences (2019, neutral): “[63] Sanchez-Lengeling, B.”
Optimization under Uncertainty in the Era of Big Data and Deep Learning: When Machine Learning Meets Mathematical Programming (2019, background): “Recently, an emerging deep generative model named generative adversarial networks (GANs) was proposed and has become increasingly popular in various areas, such as image processing [206], renewable scenario generation [207], and molecular designs [208].”
Relational Pooling for Graph Representations (2019, background): “…predict chemical properties like toxicity or photovoltaic efﬁciency (Gilmer et al., 2017; Duvenaud et al., 2015; Lee et al., 2018; Wu et al., 2018; Sanchez-Lengeling & Aspuru-Guzik, 2018); document classiﬁcation wherein a network of documents has vertices representing documents, edges…”
Generating valid Euclidean distance matrices (2019, background): “, obtain new materials [14, 15, 16, 17], design or discover pharmacological molecules such as inhibitors or antibodies [18, 19], optimize biotechnological processes [20].”
Molecular Sets (MOSES): A Benchmarking Platform for Molecular Generation Models (2018, neutral): “One important ﬁeld of application for deep generative models lies in the inverse design of organic molecules [21]: for a given functionality (solubility, ease of synthesis, toxicity or presence of particular molecular fragments, etc.) we wish to ﬁnd the perfect molecule that meets these…”
LIA: Latently Invertible Autoencoder with Adversarial Learning (2019, background): “Among emerging algorithms of novel forms due to deep learning, deep generative models play a more and more important role in cracking challenges in various scientific disciplines: high-quality image generation [15, 47, 17, 18, 4], text-to-speech transformation [41, 43], information retrieval [44], 3”
All SMILES Variational Autoencoder (2019, background): “Efﬁcient, gradient-based optimization can be performed over the space of molecules given a map between a continuous space, such as R n or the n -sphere, and the space of molecules and their properties [57].”
Inverse design of 3d molecular structures with conditional generative neural networks (2021, background): “Generative ML models have recently gained traction as a powerful, data-driven approach to inverse design as they enable sampling from a learned distribution of molecular configurations [29].”