Utilizing Parallel Biasing Methods to Understand Host-Guest Systems

Host-guest systems, where a small guest molecule resides within a pocket of a larger host molecule, are of interest in many applications, including drug delivery, toxin removal, membrane formation, hydrogels, biosensing, and many developing areas of materials science. Due to the broad variety of potential host-guest systems for a given application, molecular engineering through the computational study of host-guest binding is an attractive possibility. In this study, we explore how new “path-agnostic” algorithms, such as parallel biasing metadynamics compare with more standard “path-based” methods such as thermodynamic integration in the study of host-guest binding. This is an important question, as parallel biasing methods have particular promise for the study of binding in complicated host-guest systems, particularly those with multiple binding modes, due to their ability to efficiently explore multiple reaction coordinates simultaneously. In studying the binding affinity and other binding characteristics in systems such as cucurbit-7-uril with fentanyl, we compare our results with experiment and discuss the importance of collective variable choice in binding calculations.