Constrained Graph Variational Autoencoders for Molecule Design

By Qi Liu, Miltiadis Allamanis, Marc Brockschmidt, Alexander L. Gaunt. 2018

513 citations recorded by the source corpus.

Graphs are ubiquitous data structures for representing interactions between entities. With an emphasis on the use of graphs to represent chemical molecules, we explore the task of learning to generate graphs that conform to a distribution observed in training data. We propose a variational autoencoder model in which both encoder and decoder are graph-structured. Our decoder assumes a sequential ordering of graph extension steps and we discuss and analyze design choices that mitigate the potential downsides of this linearization. Experiments compare our approach with a wide range of baselines on the molecule generation task and show that our method is more successful at matching the statistics of the original dataset on semantically important metrics. Furthermore, we show that by using appropriate shaping of the latent space, our model allows us to design molecules that are (locally) optimal in desired properties.

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

In Materials Chemistry

In Molecular Biology

Citations in the corpus

Learning Multimodal Graph-to-Graph Translation for Molecular Optimization (2018, background): “Recently proposed graph decoders (Simonovsky & Komodakis, 2018; Li et al., 2018b; Jin et al., 2018; You et al., 2018b; Liu et al., 2018) focus on learning generative models of graphs.”
Iterative Refinement Graph Neural Network for Antibody Sequence-Structure Co-design (2021, background): “Our work is related to autoregressive models for graph generation (You et al., 2018; Li et al., 2018; Liu et al., 2018; Liao et al., 2019; Jin et al., 2020a).”
Hierarchical Generation of Molecular Graphs using Structural Motifs (2020, methodology): “1 Our baselines include state-of-the-art graph generation methods (You et al., 2018a; Liu et al., 2018; Jin et al., 2019).”
MoFlow: An Invertible Flow Model for Generating Molecular Graphs (2020, methodology): “Among the variational autoencoder (VAE)-based models [4, 5, 12, 16, 19, 20, 31], the JT-VAE [12] generates valid tree-structured molecules by first generating a tree-structured scaffold of chemical substructures and then assembling substructures according to the generated scaffold.”
All SMILES Variational Autoencoder (2019, methodology): “In contrast, molecule encoders generally use between three and seven rounds of message passing [12, 15, 27, 29, 42, 56, 70].”
Structure-aware generation of drug-like molecules (2021, methodology): “2 Training In training, the model is formulated as a variational autoencoder [31, 13].”
Self-referencing embedded strings (SELFIES): A 100% robust molecular string representation (2019, background): “Researchers have proposed many special-case solutions for overcoming these problems, (such as adapting specific machine learning models [13, 14] or changing some definitions of SMILES [15]), however, a universal solution is lacking.”
Equivariant Diffusion for Molecule Generation in 3D (2022, methodology): “Some examples are autoregressive methods such as (Liu et al., 2018; You et al., 2018; Liao et al., 2019), and one-shot approaches such as (Simonovsky & Komodakis, 2018; De Cao & Kipf, 2018; Bresson & Laurent, 2019; Kosiorek et al., 2020; Krawczuk et al., 2021).”
DiGress: Discrete Denoising diffusion for graph generation (2022, methodology): “These successes raise hope for building powerful models for graph generation, a task with diverse applications such as molecule design (Liu et al., 2018), traffic modeling (Yu & Gu, 2019), and code completion (Brockschmidt et al., 2019).”
A Gentle Introduction to Deep Learning for Graphs (2019, methodology): “encoder loss: L encoder(g) = −DKL[N(µ,σ2)kN(0,I)], (28) where DKL is the Kullback-Leibler divergence, and the two parameters of the encoding distribution are computed as µ= DGNµ(A,X) and σ= DGNσ(A,X) [112, 80, 101]. More recent approaches such as [18] propose to replace the encoder error term in Equ”
Efficient Graph Generation with Graph Recurrent Attention Networks (2019, methodology): “Examples of these models include the variational auto-encoder (VAE) model with convolutional neural networks for sequentialized molecule graphs [10] and Graph VAE models [16, 29, 23].”
Hypergraph Convolution and Hypergraph Attention (2019, background): “, 2018), chemistry (Gilmer et al., 2017; Liu et al., 2018) and NP-hard problems (Li et al.”
NeVAE: A Deep Generative Model for Molecular Graphs (2018, methodology): “…generated by several state of the art competing methods: (i) GraphVAE (Simonovsky and Komodakis 2018), (ii) GrammarVAE (Kusner et al. 2017), (iii) CVAE (Gómez-Bombarelli et al. 2016), (iv) SDVAE (Dai et al. 2018), (v) JTVAE (Jin, Barzilay, and Jaakkola 2018) and (vi) CGVAE (Liu et al. 2018).”
GuacaMol: Benchmarking Models for De Novo Molecular Design (2018, methodology): “17 This makes it diﬃcult to understand the strengths and weaknesses of the diﬀerent models, to assess which models should be used in practice, and how they can be improved and extended further.”
3D Equivariant Diffusion for Target-Aware Molecule Generation and Affinity Prediction (2023, methodology): “…molecular string representation (Bjerrum and Threlfall, 2017; Kusner et al., 2017; Segler et al., 2018) or graph representation (Li et al., 2018; Liu et al., 2018; Jin et al., 2018; Shi et al., 2020), but both representations do not take the 3D spatial interaction into account and therefore not…”