Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
80 tokens/sec
GPT-4o
59 tokens/sec
Gemini 2.5 Pro Pro
43 tokens/sec
o3 Pro
7 tokens/sec
GPT-4.1 Pro
50 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Augmenting optimization-based molecular design with graph neural networks (2312.03613v1)

Published 6 Dec 2023 in cs.CE

Abstract: Computer-aided molecular design (CAMD) studies quantitative structure-property relationships and discovers desired molecules using optimization algorithms. With the emergence of machine learning models, CAMD score functions may be replaced by various surrogates to automatically learn the structure-property relationships. Due to their outstanding performance on graph domains, graph neural networks (GNNs) have recently appeared frequently in CAMD. But using GNNs introduces new optimization challenges. This paper formulates GNNs using mixed-integer programming and then integrates this GNN formulation into the optimization and machine learning toolkit OMLT. To characterize and formulate molecules, we inherit the well-established mixed-integer optimization formulation for CAMD and propose symmetry-breaking constraints to remove symmetric solutions caused by graph isomorphism. In two case studies, we investigate fragment-based odorant molecular design with more practical requirements to test the compatibility and performance of our approaches.

PDF HTML Abstract
Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Bookmark Chat Bubble Oval Streamline Icon: https://streamlinehq.com Chat (Pro)
Definition Search Book Streamline Icon: https://streamlinehq.com
References (76)
  1. Understanding of a convolutional neural network, in: ICET.
  2. Deep learning and knowledge-based methods for computer-aided molecular design — toward a unified approach: State-of-the-art and future directions. Computers & Chemical Engineering 141, 107005.
  3. A platform of machine learning-based next-generation property estimation methods for CAMD, in: Computer Aided Chemical Engineering. volume 50, pp. 227–233.
  4. Linear model decision trees as surrogates in optimization of engineering applications. Computers & Chemical Engineering 178.
  5. Strong mixed-integer programming formulations for trained neural networks. Mathematical Programming 183, 3–39.
  6. Diffusion-convolutional neural networks, in: NeurIPS.
  7. Computer-aided molecular design: An introduction and review of tools, applications, and solution techniques. Chemical Engineering Research and Design 116, 2–26.
  8. Efficient verification of ReLU-based neural networks via dependency analysis, in: AAAI.
  9. Spectral networks and locally connected networks on graphs, in: ICLR.
  10. Branch and bound for piecewise linear neural network verification. Journal of Machine Learning Research 21.
  11. An efficient nonconvex reformulation of stagewise convex optimization problems. NeurIPS .
  12. A unified view of piecewise linear neural network verification. NeurIPS .
  13. Pyomo - Optimization Modeling in Python. volume 67. Springer.
  14. Optimization in polymer design using connectivity indices. Industrial & Engineering Chemistry Research 38, 1884–1892.
  15. OMLT: Optimization & machine learning toolkit. Journal of Machine Learning Research 23, 15829–15836.
  16. Design of polymeric membranes for air separation by combining machine learning tools with computer aided molecular design. Processes 11, 2004.
  17. Design of bio-oil solvents using multi-stage computer-aided molecular design tools, in: Computer Aided Chemical Engineering. volume 49, pp. 199–204.
  18. Novel mathematical programming model for computer aided molecular design. Industrial & Engineering Chemistry Research 35, 3788–3794.
  19. SMARTS - A language for describing molecular patterns.
  20. Convolutional neural networks on graphs with fast localized spectral filtering, in: NeurIPS.
  21. Reinforcement learning with combinatorial actions: An application to vehicle routing, in: NeurIPS.
  22. Convolutional networks on graphs for learning molecular fingerprints, in: NeurIPS.
  23. Deep learning for molecular design — a review of the state of the art. Molecular Systems Design & Engineering 4, 828–849.
  24. Deep graph generators: A survey. IEEE Access 9, 106675–106702.
  25. Fast graph representation learning with PyTorch Geometric, in: ICLR 2019 Workshop on Representation Learning on Graphs and Manifolds.
  26. Deep neural networks and mixed integer linear optimization. Constraints 23, 296–309.
  27. Fundamental domains for integer programs with symmetries, in: Combinatorial Optimization and Applications.
  28. Computer-aided methods and tools for chemical product design. Chemical Engineering Research and Design 82, 1494–1504.
  29. Editorial overview: Frontiers of chemical engineering: chemical product design II. Current Opinion in Chemical Engineering 35, 100783.
  30. Sample efficiency matters: A benchmark for practical molecular optimization, in: NeurIPS Track Datasets and Benchmarks.
  31. Neural message passing for quantum chemistry, in: ICML.
  32. DeepFrag: a deep convolutional neural network for fragment-based lead optimization. Chemical Science 12, 8036–8047.
  33. Gurobi Optimizer Reference Manual. URL: https://www.gurobi.com.
  34. Inductive representation learning on large graphs, in: NeurIPS.
  35. Design of mosquito repellent molecules via the integration of hyperbox machine learning and computer aided molecular design. Digital Chemical Engineering 3, 100018.
  36. Polytopes associated with symmetry handling. Mathematical Programming 175, 197–240.
  37. Deep learning volatility: A deep neural network perspective on pricing and calibration in (rough) volatility models. Quantitative Finance 21, 11–27.
  38. When deep learning meets polyhedral theory: A survey. arXiv preprint arXiv:2305.00241 .
  39. Junction tree variational autoencoder for molecular graph generation, in: ICML.
  40. Hierarchical generation of molecular graphs using structural motifs, in: ICML.
  41. Semi-supervised classification with graph convolutional networks, in: ICLR.
  42. OptCAMD: An optimization-based framework and tool for molecular and mixture product design. Computers & Chemical Engineering 124, 285–301.
  43. Group contribution-based property modeling for chemical product design: A perspective in the AI era. Fluid Phase Equilibria 568, 113734.
  44. Mixed-integer optimisation of graph neural networks for computer-aided molecular design. arXiv preprint arXiv:2312.01228 .
  45. Optimization of tree ensembles. Operations Research 68, 1605–1624.
  46. Mixed-integer convex nonlinear optimization with gradient-boosted trees embedded. INFORMS Journal on Computing 33, 1103–1119.
  47. Challenges and opportunities in computer aided molecular design. Computer Aided Chemical Engineering 34, 25–34.
  48. Learning convolutional neural networks for graphs, in: ICML.
  49. Computer aided molecular design: A novel method for optimal solvent selection. Fluid Phase Equilibria 82, 47–54.
  50. Constrained discrete black-box optimization using mixed-integer programming, in: ICML.
  51. A deep generative model for fragment-based molecule generation, in: AISTATS.
  52. Fragment-based ligand generation guided by geometric deep learning on protein-ligand structures, in: ICLR 2022 Workshop MLDD.
  53. Graph machine learning for design of high-octane fuels. AIChE Journal , e17971.
  54. CAQL: Continuous action Q-learning, in: ICLR.
  55. Design of alternative refrigerants via global optimization. AIChE Journal 49, 1761–1775.
  56. Nonlinear hybrid planning with deep net learned transition models and mixed-integer linear programming, in: IJCAI.
  57. Graph neural networks for prediction of fuel ignition quality. Energy & Fuels 34, 11395–11407.
  58. Scaling up exact neural network compression by ReLU stability, in: NeurIPS.
  59. SMILES to smell: decoding the structure–odor relationship of chemical compounds using the deep neural network approach. Journal of Chemical Information and Modeling 61, 676–688.
  60. Gated graph recursive neural networks for molecular property prediction. arXiv preprint arXiv:1909.00259 .
  61. Environmentally benign solvent design by global optimization. Computers & Chemical Engineering 23, 1381–1394.
  62. Intriguing properties of neural networks, in: ICLR.
  63. Entmoot: A framework for optimization over ensemble tree models. Computers & Chemical Engineering 151, 107343.
  64. A fragrance prediction model for molecules using rough set-based machine learning. Chemie Ingenieur Technik 95, 438–446.
  65. Evaluating robustness of neural networks with mixed integer programming, in: ICLR.
  66. Partition-based formulations for mixed-integer optimization of trained ReLU neural networks, in: NeurIPS.
  67. Optimizing over an ensemble of trained neural networks. INFORMS Journal on Computing .
  68. Molecule property prediction based on spatial graph embedding. Journal of Chemical Information and Modeling 59, 3817–3828.
  69. Building attention and edge message passing neural networks for bioactivity and physical–chemical property prediction. Journal of Cheminformatics 12, 1–18.
  70. Scalable planning with deep neural network learned transition models. Journal of Artificial Intelligence Research 68, 571–606.
  71. Graph-based generative models for de Novo drug design. Drug Discovery Today: Technologies 32, 45–53.
  72. Graph neural networks for automated de novo drug design. Drug Discovery Today 26, 1382–1393.
  73. Deep learning based regression and multiclass models for acute oral toxicity prediction with automatic chemical feature extraction. Journal of Chemical Information and Modeling 57, 2672–2685.
  74. Analyzing learned molecular representations for property prediction. Journal of Chemical Information and Modeling 59, 3370–3388.
  75. Generic mathematical programming formulation and solution for computer-aided molecular design. Computers & Chemical Engineering 78, 79–84.
  76. Optimizing over trained GNNs via symmetry breaking, in: NeurIPS.
User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (6)
  1. Shiqiang Zhang (18 papers)
  2. Juan S. Campos (5 papers)
  3. Christian Feldmann (5 papers)
  4. Frederik Sandfort (3 papers)
  5. Miriam Mathea (5 papers)
  6. Ruth Misener (43 papers)
Citations (4)
View on Semantic Scholar