ChemTS: an efficient python library for de novo molecular generation

By Xiufeng Yang, Jinzhe Zhang, Kazuki Yoshizoe, Kei Terayama, Koji Tsuda. 2017

257 citations recorded by the source corpus.

Abstract Automatic design of organic materials requires black-box optimization in a vast chemical space. In conventional molecular design algorithms, a molecule is built as a combination of predetermined fragments. Recently, deep neural network models such as variational autoencoders and recurrent neural networks (RNNs) are shown to be effective in de novo design of molecules without any predetermined fragments. This paper presents a novel Python library ChemTS that explores the chemical space by combining Monte Carlo tree search and an RNN. In a benchmarking problem of optimizing the octanol-water partition coefficient and synthesizability, our algorithm showed superior efficiency in finding high-scoring molecules. ChemTS is available at https://github.com/tsudalab/ChemTS. Graphical Abstract

Subfields: Computational Theory and Mathematics, Materials Chemistry, Molecular Biology

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

1. Jure Leskovec
2. Sepp Hochreiter
3. Thomas Unterthiner
4. Günter Klambauer
5. Vijay S. Pande
6. Bowen Liu
7. Rex Ying
8. Tianfan Fu

In Materials Chemistry

1. Sepp Hochreiter
2. Jure Leskovec
3. Thomas Unterthiner
4. Günter Klambauer
5. Philipp Renz
6. Kristina Preuer
7. Vijay S. Pande
8. Kevin Maik Jablonka

In Molecular Biology

1. Rex Ying
2. Vijay S. Pande
3. Jiaxuan You
4. Sepp Hochreiter
5. Thomas Unterthiner
6. Jure Leskovec
7. Bowen Liu
8. Günter Klambauer

Citations in the corpus

• Graph Convolutional Policy Network for Goal-Directed Molecular Graph Generation (2018, background): “Recently, there has been significant advances in applying deep learning models to molecule generation [15, 38, 7, 9, 22, 4, 31, 27, 34, 42].”
• Automatic Chemical Design Using a Data-Driven Continuous Representation of Molecules (2016, neutral): “− 47 Several proceeding works have further explored the use of Long Short-Term Memory (LSTM) networks and recurrent networks applied to SMILES strings to generate new molecules 48,49 and predict the outcomes of organic chemistry reactions.”
• In silico generation of novel, drug-like chemical matter using the LSTM neural network (2017, methodology): “[13] combined a Monte Carlo tree search and a recurrent neural network to generate molecules with desired values of octanol-water partition coefficient and good synthesizability.”
• The Role of Machine Learning in the Understanding and Design of Materials (2020, background): “116 A desired property to circumvent the expensive optimization in discrete, enormous chemical space is to develop continuous and di ff erentiable representations of chemical structures.”
• Fréchet ChemNet Distance: A Metric for Generative Models for Molecules in Drug Discovery (2018, background): “Other metrics focus on numerical or visual comparison of molecular properties like solubility [17], number of rotational bonds [17], number of aromatic rings [17], synthetic accessibility [8, 12, 28] and druglikeness [8, 12].”
• MolGenSurvey: A Systematic Survey in Machine Learning Models for Molecule Design (2022, background): “ChemTS [149] None SMILES MCTS+AR MP-MCTS [114] None SMILES MCTS+AR”
• Fréchet ChemblNet Distance: A metric for generative models for molecules (2018, background): “Other metrics focus on numerical or visual comparison of molecular properties like solubility [15], number of rotational bonds [15], number of aromatic rings [15], synthetic accessibility [8, 11, 23] and druglikeness [8, 11].”