Jethro Akroyd

The discovery of functional reticular materials is dependent on optimising the organic building unit to achieve precise chemical and structural properties. However, experimentally characterised chemical building units (CBUs) are only a small fraction of the possible design space, and as such, computer-aided design based on datasets of known reticular materials have fundamental limitations in chemical diversity. Here, we present a general workflow for extracting, analysing, and recombining molecular fragments from large, chemically diverse datasets to systematically expand the accessible organic CBU design space. Using the ChEMBL36 database as a case study, we generate a library of 12 387 unique fragments through filtering, fragmentation, and ontology-driven classification. These fragments are recombined to enumerate a dataset of 44.8 million CBUs. To demonstrate integrating the workflow into designing functional reticular materials, we combine the library with a genetic algorithm to identify candidate metal-organic polyhedra (MOPs) carriers for the antibiotic ciprofloxacin, with the highest-ranked candidates predicted to have low toxicity and stable host-guest interactions according to tight-binding density functional theory calculations. The concepts and relationships between the molecular fragments, CBUs, templates, and MOPs are captured and connected through a knowledge graph and ontology integrated in The World Avatar.

Reticular materials have come to the fore of chemistry with exceptional potential in applications ranging from CO₂ capture and chemical separations to catalysis and drug delivery. However, due to the vast combinatorial space of molecular building blocks that can form these materials, designing high-performing reticular materials for applications remains a considerable challenge. Here, we present a computational approach that combines a library of molecular fragments suitable for constructing organic building units, template-based reassembly, and evolutionary optimization to accelerate the discovery of reticular materials. Applied to metal-organic polyhedra (MOPs), this approach produces a design space of nearly 800,000 MOP configurations. A genetic algorithm (GA) based on the molecular fragments is shown to be effective at rapidly identifying optimal MOPs within this space, demonstrated through optimizing cavity properties for host-guest applications and CO₂ interaction energies estimated by machine-learning-accelerated simulations. An important component of our approach is that it is fully ontologized and integrated within The World Avatar, forming part of a broader, interoperable knowledge model for the discovery of reticular materials.

Methods are benchmarked for refining in silico assembled metal–organic polyhedra and the results integrated into a knowledge graph for high-throughput screening.

Assembly modeling has been achieved in knowledge AI systems for the automated inference of new and rational metal–organic polyhedra (J. Am. Chem. Soc. 2022, 144, 26, 11713–11728). This work presents an algorithm and data structure that extends the process of assembly modeling to the automated generation of structural information about metal–organic polyhedra, enabling automation of computational approaches to analyse trends in cavity and pore sizing. Distinct from string‐based tools for purely organic cages, the workflow positions inorganic, organic, and hybrid chemical building units directly in 3‐D and outputs geometries suitable for higher‐level geometry optimization calculations in one step. The structural geometries obtained from this work are semantically integrated as part of The World Avatar, a dynamic knowledge ecosystem.

This work presents a generalisable process that transforms unstructured synthesis descriptions of metal–organic polyhedra into machine-readable, structured representations, integrating them into The World Avatar, a universal knowledge framework.

The open-source Python package ‘twa’ streamlines knowledge-graph-driven chemical data integration and workflow automation, unifying fragmented data and expanding the OntoMOPs knowledge graph with 799 new MOPs. It is available via ‘pip install twa’.