Protein Folding with the Parallel Replica Exchange Molecular Dynamics Method - Parallel Computing for Bioinformatics and Computational Biology - page 397

CHAPTER 17
Protein Folding with the Parallel
Replica Exchange Molecular
Dynamics Method
RUHONG ZHOU
A highly parallel replica exchange molecular dynamics (REMD) method and its
application in protein folding and protein structure prediction are described in this
chapter. The REMD method couples molecular dynamics trajectories with a tem-
perature exchange Monte Carlo process for efficient sampling of the conformational
space. Two example protein systems, one α -helix and one β -hairpin, are used to
demonstrate the power of the algorithm. Up to 64 replicas of solvated protein
systems are simulated in parallel over a wide range of temperatures. Very high effi-
ciency ( > 98%) can be achieved with this embarrassingly parallel algorithm. The
simulation results show that the combined trajectories in temperature and config-
urational space allow a replica to overcome free energy barriers present at low
temperatures. These large-scale simulations also reveal detailed results on folding
mechanisms, intermediate state structures, thermodynamic properties, and the tem-
perature dependences for both protein systems. Furthermore, the extensive data from
REMD simulations are used to assess the various solvation models and force fields,
which provides insights into the fix of the problems and further improvement of the
models. Finally, the usage of the REMDmethod in protein structure refinement is also
discussed.
17.1 INTRODUCTION
How to efficiently sample the conformational space of complex biological systems,
such as protein folding, remains a great challenge inmolecular biology [1-6]. The free
energy landscape of protein folding is believed to be at least partially rugged. At room