Graph Theory-Based De Novo Strategy for Automated Generation of Organic Molecular Cage
Year: 2026
Type: preprint
Abstract: The current design and construction of organic molecular cages (OMCs) remain largely constrained by established topological library. To address this challenge, we established a de novo, graph-theory-based strategy that enables the automated generation of OMCs architectures, which require only the chemical structures of building blocks. By decomposing OMCs into distinct topological geometric elements, a system of equations can be established using graph theory to describe the mathematical relationships and geometric constraints for their architectures. Solving these equations yields the numerical composition of the topological geometric elements for each candidate OMC, allowing the potential rotational symmetries of their architectures to be inferred. Based on the inferred rotational symmetries, a database of OMC topologies was systematically established by applying symmetry operations to define the connectivity between geometric elements. Subsequently, different molecular building blocks were incorporated into these topologies, followed by geometry optimization to generate a large space of chemically realistic three-dimensional OMC structures. This strategy enabled the generation of OMCs with previously unexplored topologies, substantially expanding the structural diversity of the OMCs’ chemical space. To further enrich this chemical space, we extended our strategy to encompass isomers arising from mixed or prochiral building blocks. This strategy enables the systematic exploration of previously overlooked OMC topologies and their corresponding isomers, providing a foundation for the targeted design of molecular cages with tailored cavity geometries, pore sizes, and chemical environments for applications in molecular recognition, selective separation and catalysis.
Source: ChemRxiv
Language: English
Subfield: Organic Chemistry
Field: Chemistry
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7222229843