The relevance of vibrational modes in flexible docking of proteins

Protein-protein interactions (PPI) are fundamental to many biological processes, and predicting their binding conformations is critical for understanding their function. When two proteins bind without undergoing conformational changes, rigid-body docking combined with improved scoring functions can predict PPI structure. However, many unbound monomers undergo significant conformational changes upon binding, which makes PPI structure prediction more difficult. Understanding the path that the monomers take from the unbound to the bound conformation is key to predicting the correct PPI structure. It has been hypothesized that the low-frequency vibrational modes in proteins may provide information about the pathway from the unbound to bound conformation. In this work, we carry out molecular dynamics simulations of the individual monomers in PPIs starting from the relaxed, unbound conformation. We first determine whether the vibrational modes for folded proteins are well-defined by comparing the eigenvalues of the Hessian in the relaxed, unbound conformation to the modes obtained from the Fourier transform of the velocity autocorrelation function starting from the unbound conformation as a function temperature relative to the folding temperature. Assuming that the vibrational modes are robust, we will then determine whether the change from the unbound to the bound conformation can be predicted by moving the system along the low-frequency modes of the Hessian.