Extension of the Kinematics-Based Method for Predicting the Motion of Proteins
On the basis of an analogy between the kinematic structures of proteins and robotic mechanisms, we have so far developed methods for predicting the internal motion of proteins from three-dimensional structural data in the protein data bank (PDB). With these methods, we model proteins as serial manipulators constrained by springs, and calculate the structural compliance of the protein model. In this study, toward more practical purposes, we reformulate and extend the existing methods by broadening the definition of structural compliance and reducing the number of variables for expressing the conformation of the model. The broadening is performed by separating the parts whose deformations are evaluated from those where forces are applied. This separation allows the calculation of the effective forces causing deformation in other specified parts. We also reduce the number of conformation variables from the consideration based on the algebraic structure of the basic equations. The size of the matrix whose inverse must be calculated is thus minimized, and the computational cost is reduced. We verify the effectiveness of these extensions by analyzing the PDB data of some proteins.