CrypWater: Detection and Characterization of Protein Cryptic Pocket through Water Density Fluctuation in Molecular Dynamics Simulation

Tianming Qu, Lianqing Zheng, Wei Yang

Florida State University, Department of Chemistry

Employing molecular dynamics simulation methods to identify novel protein cryptic pockets has attracted increasing interest for drug discovery. A challenge has been on how to detect and quantify druggable site formation from a molecular dynamics (MD) trajectory robustly and accurately. In the past decades, several strategies have been developed. Even having been available for public usage, most of them still suffer from lack of either robustness or accuracy; and so, in practice, prior human input still plays a dominant role in their application for cryptic pocket characterization. In this paper, we are reporting a new tool, CrypWater. This method uniquely utilizes water density fluctuation and the convex hull feature of auto-identified structure transition regions to detect and characterize cryptic pockets. To demonstrate this CrypWater method, ERK2 kinase, our MD study of which led to visible cryptic pocket generation, was chosen as the model system. In addition, several commonly known protein pocket detection methods were comparatively investigated. As revealed from this model study, CrypWater significantly outperformed all the other existing approaches. It not only accurately detected and characterized the new cryptic pocket but also for the first time truly satisfied the no-human-input identification requirement. Furthermore, derived water density information is readily employed for druggability analysis.