Premium
Domain decomposition‐based structural condensation of large protein structures for understanding their conformational dynamics
Author(s) -
Kim Jae In,
Na Sungsoo,
Eom Kilho
Publication year - 2010
Publication title -
journal of computational chemistry
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.907
H-Index - 188
eISSN - 1096-987X
pISSN - 0192-8651
DOI - 10.1002/jcc.21613
Subject(s) - granularity , biological system , normal mode , hessian matrix , computer science , computation , molecular dynamics , frequency domain , protein structure , domain decomposition methods , statistical physics , algorithm , chemistry , computational chemistry , physics , mathematics , thermodynamics , nuclear magnetic resonance , vibration , quantum mechanics , finite element method , computer vision , biology , operating system
Normal mode analysis (NMA) with coarse‐grained model, such as elastic network model (ENM), has allowed the quantitative understanding of protein dynamics. As the protein size is increased, there emerges the expensive computational process to find the dynamically important low‐frequency normal modes due to diagonalization of massive Hessian matrix. In this study, we have provided the domain decomposition‐based structural condensation method that enables the efficient computations on low‐frequency motions. Specifically, our coarse‐graining method is established by coupling between model condensation (MC; Eom et al., J Comput Chem 2007, 28 , 1400) and component mode synthesis (Kim et al., J Chem Theor Comput 2009, 5 , 1931). A protein structure is first decomposed into substructural units, and then each substructural unit is coarse‐grained by MC. Once the NMA is implemented to coarse‐grained substructural units, normal modes and natural frequencies for each coarse‐grained substructural unit are assembled by using geometric constraints to provide the normal modes and natural frequencies for whole protein structure. It is shown that our coarse‐graining method enhances the computational efficiency for analysis of large protein complexes. It is clearly suggested that our coarse‐graining method provides the B‐factors of 100 large proteins, quantitatively comparable with those obtained from original NMA, with computational efficiency. Moreover, the collective behaviors and/or the correlated motions for model proteins are well delineated by our suggested coarse‐grained models, quantitatively comparable with those computed from original NMA. It is implied that our coarse‐grained method enables the computationally efficient studies on conformational dynamics of large protein complex. © 2010 Wiley Periodicals, Inc. J Comput Chem, 2010