Mixed levels of coarse-graining of large proteins using elastic network model succeeds in extracting the slowest motions

We perform a mixed coarse-graining approach in a normal mode analysis of protein motions, which enables the modeling of a protein's native conformation with different regions having low and high resolution. As a result, the dynamics of the interesting functional parts within a supramolecular assemblage can be analyzed at high resolution, while the remainder of the structure is represented at poorer resolution, thus keeping the total number of nodes in the system sufficiently low for computational tractability. Our results indicate that the vibrational dynamics of specific components in a large multi-subunit protein are best described by retaining all the components of the structure, whether at higher or lower resolution. It is also shown that similar frequency distributions are obtained for different proteins and at different levels of coarse-graining, at the lower end of the spectrum, where the most significant slowest motions occur. (C) 2003 Elsevier Ltd. All rights reserved.