mlsanityData to 7 Oct 2026

The power of motifs as inductive bias for learning molecular distributions

arXiv 2306.17246 ↗First posted 4 Apr 2023cs.LG q-bio.BM stat.APcited by 6

Not on the map. Why?

Abstract (arXiv)

Machine learning for molecules holds great potential for efficiently exploring the vast chemical space and thus streamlining the drug discovery process by facilitating the design of new therapeutic molecules. Deep generative models have shown promising results for molecule generation, but the benefits of specific inductive biases for learning distributions over small graphs are unclear. Our study aims to investigate the impact of subgraph structures and vocabulary design on distribution learning, using small drug molecules as a case study. To this end, we introduce Subcover, a new subgraph-based fragmentation scheme, and evaluate it through a two-step variational auto-encoder. Our results show that Subcover's improved identification of chemically meaningful subgraphs leads to a relative improvement of the FCD score by 30%, outperforming previous methods. Our findings highlight the potential of Subcover to enhance the performance and scalability of existing methods, contributing to the advancement of drug discovery.

Key facts. Cited by 6 papers in the map (data through 7 Oct 2026). Builds on 38 of its 51 references that are in the map. First posted on arXiv 4 Apr 2023.

Builds on

38 of 51 references.

References · 38

  1. 2016Automatic chemical design using a data-driven continuous representation of moleculescited by 762
  2. 2018Junction Tree Variational Autoencoder for Molecular Graph Generationcited by 647
  3. 2020Masked Label Prediction: Unified Message Passing Model for Semi-Supervised Classificationcited by 500
  4. 2018MolGAN: An implicit generative model for small molecular graphscited by 392
  5. 2021Flow Network based Generative Models for Non-Iterative Diverse Candidate Generationcited by 384
  6. 2018GraphVAE: Towards Generation of Small Graphs Using Variational Autoencoderscited by 375
  7. 2018Graph Convolutional Policy Network for Goal-Directed Molecular Graph Generationcited by 363
  8. 2020GraphAF: a Flow-based Autoregressive Model for Molecular Graph Generationcited by 301
  9. 2018GuacaMol: Benchmarking Models for De Novo Molecular Designcited by 294
  10. 2017Grammar Variational Autoencodercited by 283
  11. 2018Molecular Sets (MOSES): A Benchmarking Platform for Molecular Generation Modelscited by 270
  12. 2018Learning Deep Generative Models of Graphscited by 227
26 more
  1. 2020Hierarchical Generation of Molecular Graphs using Structural Motifscited by 204
  2. 2018Fr'echet ChemNet Distance: A metric for generative models for molecules in drug discoverycited by 191
  3. 2020MoFlow: An Invertible Flow Model for Generating Molecular Graphscited by 183
  4. 2018Constrained Graph Variational Autoencoders for Molecule Designcited by 182
  5. 2021GraphDF: A Discrete Flow Model for Molecular Graph Generationcited by 177
  6. 2022Sample Efficiency Matters: A Benchmark for Practical Molecular Optimizationcited by 151
  7. 2021GraphEBM: Molecular Graph Generation with Energy-Based Modelscited by 85
  8. 2019MolecularRNN: Generating realistic molecular graphs with optimized propertiescited by 80
  9. 2018Constrained Generation of Semantically Valid Graphs via Regularizing Variational Autoencoderscited by 74
  10. 2021Learning to Extend Molecular Scaffolds with Structural Motifscited by 70
  11. 2022A Survey on Deep Graph Generation: Methods and Applicationscited by 69
  12. 2021Hit and Lead Discovery with Explorative RL and Fragment-based Molecule Generationcited by 68
  13. 2018NeVAE: A Deep Generative Model for Molecular Graphscited by 66
  14. 2019Augmenting Genetic Algorithms with Deep Neural Networks for Exploring the Chemical Spacecited by 66
  15. 2022MolGenSurvey: A Systematic Survey in Machine Learning Models for Molecule Designcited by 66
  16. 2018Molecular Hypergraph Grammar with its Application to Molecular Optimizationcited by 56
  17. 2021Molecule Generation by Principal Subgraph Mining and Assemblingcited by 53
  18. 2022Predicting Cellular Responses to Novel Drug Perturbations at a Single-Cell Resolutioncited by 42
  19. 2019A Two-Step Graph Convolutional Decoder for Molecule Generationcited by 25
  20. 2018DEFactor: Differentiable Edge Factorization-based Probabilistic Graph Generationcited by 24
  21. 2022Molecular Representation Learning via Heterogeneous Motif Graph Neural Networkscited by 23
  22. 2017ChemGAN challenge for drug discovery: can AI reproduce natural chemical diversity?cited by 21
  23. 2020Generative chemistry: drug discovery with deep learning generative modelscited by 21
  24. 2020Graph Deconvolutional Generationcited by 16
  25. 2022Molecule Generation for Drug Design: a Graph Learning Perspectivecited by 13
  26. 2019Scaffold-based molecular design using graph generative modelcited by 6

Built on by

Cited by year

Citing papers per year: 2023: 1; 2024: 3; 2025: 2.

6 papers.

  1. 2024Less can be more for predicting properties with large language modelscited by 10
  2. 2024Expressivity and Generalization: Fragment-Biases for Molecular GNNscited by 8
  3. 2023MAGNet: Motif-Agnostic Generation of Molecules from Shapescited by 7
  4. 2025Unified Guidance for Geometry-Conditioned Molecular Generationcited by 7
  5. 2025Byte Pair Encoding for Efficient Time Series Forecastingcited by 5
  6. 2024A Review on Fragment-based De Novo 2D Molecule Generationcited by 3