Hierarchical generative modeling for the design of multi-component systems
Abstract
The functionality of catalysts, enzymes, and supramolecular assemblies emerges not from individual molecules alone, but from the subtle interplay between multiple components arranged in complex systems. Designing such systems is a grand challenge: the combinatorial explosion of possible chemical compositions and spatial arrangements makes brute-force exploration infeasible, while many current generative approaches remain limited to isolated molecules. In this work, we introduce a hierarchical generative optimization framework that overcomes this barrier by coupling a genetic algorithm for configurational search with a generative model for molecular design. This closed-loop approach enables simultaneous refinement of geometry and composition, efficiently steering discovery toward systems with targeted functionality. As a proof of concept, we design catalytic environments for the Claisen rearrangement of p-tolyl ether by optimizing surrounding components around a fixed reference transition-state geometry. Despite this constraint during the search phase, post-hoc validation via Climbing-Image Nudged Elastic Band calculations, with the environment held fixed, indicates a reduction of the activation barrier by about 30%. Beyond this example, the framework applies to systems in which a well-defined central structure is stabilized by non-covalently interacting components, such as supramolecular hosts and local catalytic environments.
Related
- HiPoly: a hierarchical polymer-native AI framework for property prediction and generative design
- Hierarchical Surrogate Modeling and Optimization of the SO2 Recovery–Water Loss Trade-Off in Basic Aluminum Sulfate Regeneration
- GEOMIND: a hybrid generative artificial intelligence model for geopolymer design and optimization
- Transformative Powder Fibration toward Hierarchical Ceramic Aerogels for Multifunctional Aerospace Systems
- Slip casting and stress-controlled co-sintering of a complex ceramic Al₂O₃–ZrO₂/Al₂O₃ bi-material component through bevelled interface design for electromagnetic applications
BibTeX
@article{barrett2026hierarchical,
title = {{Hierarchical generative modeling for the design of multi-component systems}},
author = {Rhyan Barrett and Robin Curth and Julia Westermayr},
journal = {Journal of Cheminformatics},
year = {2026},
month = oct,
doi = {10.1186/s13321-026-01313-3},
url = {https://link.springer.com/10.1186/s13321-026-01313-3},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Hierarchical generative modeling for the design of multi-component systems
AU - Rhyan Barrett
AU - Robin Curth
AU - Julia Westermayr
PY - 2026
DA - 2026/10/11
JO - Journal of Cheminformatics
DO - 10.1186/s13321-026-01313-3
UR - https://link.springer.com/10.1186/s13321-026-01313-3
AB - The functionality of catalysts, enzymes, and supramolecular assemblies emerges not from individual molecules alone, but from the subtle interplay between multiple components arranged in complex systems. Designing such systems is a grand challenge: the combinatorial explosion of possible chemical compositions and spatial arrangements makes brute-force exploration infeasible, while many current generative approaches remain limited to isolated molecules. In this work, we introduce a hierarchical generative optimization framework that overcomes this barrier by coupling a genetic algorithm for configurational search with a generative model for molecular design. This closed-loop approach enables simultaneous refinement of geometry and composition, efficiently steering discovery toward systems with targeted functionality. As a proof of concept, we design catalytic environments for the Claisen rearrangement of p-tolyl ether by optimizing surrounding components around a fixed reference transition-state geometry. Despite this constraint during the search phase, post-hoc validation via Climbing-Image Nudged Elastic Band calculations, with the environment held fixed, indicates a reduction of the activation barrier by about 30%. Beyond this example, the framework applies to systems in which a well-defined central structure is stabilized by non-covalently interacting components, such as supramolecular hosts and local catalytic environments.
SN - 1758-2946
ER -