Book of Abstracts: Albany 2003
June 17-21 2003
A Comparison Between Elastic Network Interpolation and MD Simulation of 16S Ribosomal RNA
Molecular dynamics (MD) simulation is a useful tool to analyze the evolution of macromolecular conformations due to the influence of internal and external forces. In MD, trajectories are calculated by the classical Newtonian equations of motion sometimes using X-ray crystallography data as the initial atomic coordinates. MD simulation provides realistic molecular motions including the effect of surrounding solvent. However, the computational cost is very high and it is very difficult to obtain long-time-scale collective motions from even millions of MD timesteps. Normal mode analysis (NMA) has been used to study slow and collective behaviors of macromolecules around low energy equilibrium conformations. It is much more computationally efficient than MD simulation but has the limitation of substantial data storage requirements in the case of large macromolecules. In addition, NMA is limited in its ability to predict anharmonic motions and pathways. To circumvent these limitations of both MD simulation and NMA, we have investigated the performance of a coarse-grained elastic network interpolation model. MD simulation for the core central domain of the 16S Ribosomal RNA (16S rRNA) was performed and sampled at 5000 conformations. Using elastic network interpolation, we generated a 1 degree-of-freedom (DOF) pathway over the 10Å swing between the two extreme conformations which occurred during the MD run and then did coarse-grained NMA with a simplified harmonic potential for each of the intermediates. By parameterizing fluctuations about the 1 DOF elastic network pathway using only 1% of the normal modes (about 40 DOF) we were able to capture all 5000 conformations of the MD run as fluctuations around our pathway (with 2Å RMS deviation on average). In addition, if we restrict attention to only those MD conformations that occur between the times of the two extreme conformations, RMSD between our pathway with superimposed 1% normal modes is about 1Å away from the MD results. These results may serve as a paradigm for reduced-degree-of-freedom dynamic simulations of large biological macromolecules as well as a method for the reduced-parameter interpretation of massive amounts of MD data.
Moon K. Kim1
1Department of Mechanical Engineering