Search for: All records

Identifying informative low-dimensional features that characterize dynamics in molecular simulations remains a challenge, often requiring extensive manual tuning and system-specific knowledge. Here, we introduce geom2vec, in which pretrained graph neural networks (GNNs) are used as universal geometric featurizers. By pretraining equivariant GNNs on a large dataset of molecular conformations with a self-supervised denoising objective, we obtain transferable structural representations that are useful for learning conformational dynamics without further fine-tuning. We show how the learned GNN representations can capture interpretable relationships between structural units (tokens) by combining them with expressive token mixers. Importantly, decoupling training the GNNs from training for downstream tasks enables analysis of larger molecular graphs (that can represent small proteins at all-atom resolution) with limited computational resources. In these ways, geom2vec eliminates the need for manual feature selection and increases the robustness of simulation analyses. 

Abstract An anionic Rh−Ga complex catalyzed the hydrodefluorination of challenging C−F bonds in electron‐rich aryl fluorides and trifluoromethylarenes when irradiated with violet light in the presence of H₂, a stoichiometric alkoxide base, and a crown‐ether additive. Based on theoretical calculations, the lowest unoccupied molecular orbital (LUMO), which is delocalized across both the Rh and Ga atoms, becomes singly occupied upon excitation, thereby poising the Rh−Ga complex for photoinduced single‐electron transfer (SET). Stoichiometric and control reactions support that the C−F activation is mediated by the excited anionic Rh−Ga complex. After SET, the proposed neutral Rh⁰intermediate was detected by EPR spectroscopy, which matched the spectrum of an independently synthesized sample. Deuterium‐labeling studies corroborate the generation of aryl radicals during catalysis and their subsequent hydrogen‐atom abstraction from the THF solvent to generate the hydrodefluorinated arene products. Altogether, the combined experimental and theoretical data support an unconventional bimetallic excitation that achieves the activation of strong C−F bonds and uses H₂and base as the terminal reductant.