Identification of slow relaxation modes in a protein trimer via positive definite relaxation mode analysis.

Recently, dynamic analysis methods in signal processing have been applied to the analysis of molecular dynamics (MD) trajectories of biopolymers. In the context of a relaxation mode analysis (RMA) method, based on statistical physics, it is explained why the signal-processing methods work well for the simulation trajectories of biopolymers. A distinctive difference between the RMA method and the signal-processing methods is the introduction of an additional parameter, called an evolution time parameter. This parameter enables us to better estimate the relaxation modes and rates, although it increases computational difficulty. In this paper, we propose a simple and effective extension of the RMA method, which is referred to as the positive definite RMA method, to introduce the evolution time parameter robustly. In this method, an eigenvalue problem for the time correlation matrix of physical quantities relevant to slow relaxation in a system is first solved to find the subspace in which the matrix is numerically positive definite. Then, we implement the RMA method in the subspace. We apply the method to the analysis of a 3-μs MD trajectory of a heterotrimer of an erythropoietin protein and two of its receptor proteins, and we demonstrate the effectiveness of the method.