Accommodating Statistical and Physics-Based Computational Protocols for Molecular Glue Model and Evaluation.

Molecular glues (MGs) offer a promising strategy for modulating protein-protein interactions (PPIs) by inducing novel intermolecular contacts or stabilizing existing ones. However, despite cryo-electron microscopy and related techniques providing high-resolution structural models, their substantial resource requirements─including specialized equipment, expertise, and processing time─often restrict their application in the rational design of similar MGs. Recognizing the diverse mechanisms of MG action, we employ two distinct computational approaches: AlphaFold-Multimer and molecular docking, tailored to three specific MG systems. By validating these models against experimental crystal structures and elucidating the dynamic mechanisms underlying molecular glue formation, we establish a foundation for developing more effective compounds. Furthermore, molecular dynamics simulations provide atomic-resolution snapshots of the ternary complexes over time, including water-mediated interaction networks, revealing dynamic information that can guide the design of molecular glues with distinct kinetic profiles or specificity properties influenced by peripheral molecular events. Our findings suggest that structure-based computational approaches will be increasingly pivotal in rationalizing these therapeutically promising MGs.