Manifold Diffusion for Structure Generation of Transition Metal Complexes
This work addresses the challenging problem of structure generation for transition metal complexes, which is critical for catalysis and materials design, but the method is domain-specific and incremental.
TMCgen, a manifold diffusion model, generates accurate 3D geometries of transition metal complexes by focusing on coordination angles and ligand torsions, outperforming existing methods in efficiency and accuracy on experimental datasets.
Transition metal complexes are central to catalysis, drug design, and materials science, with relevant properties strongly sensitive to their three-dimensional geometry. However, the electronic diversity and unconventional bonding environments of transition metal complexes pose a major challenge for accurate structure generation. In this work, we introduce TMCgen, a manifold diffusion machine learning model that efficiently and accurately generates geometries of transition metal complexes. By formulating the diffusion process over the metal-ligand coordination angles, combined with torsional and rotational diffusion of the ligands, TMCgen focuses on the key geometric degrees of freedom of transition metal complexes. TMCgen shows strong performance in generating accurate coordination environments on a diverse set of experimentally derived bioinorganic and organometallic complexes while requiring only few inference steps, enabling efficient generation. Our results demonstrate the potential of manifold-based generative modeling for data-efficient geometry generation, paving the way for property-conditioned design of transition metal complexes.