Developing effective parallel programs and algorithms for all-atom protein models containing Monte Carlo simulati̇on technique

For better understanding the dynamics of proteins, it is aimed to propose new all-atom Monte Carlo protein models enabling us to investigate the fundamental physical and chemical principals of interactions among atoms. We have parallelized our Monte Carlo simulation program for protein specific Go model approach by using the Grand Central Dispatch library. The results indicate that using well potentials and Lennard-Jones potentials show up to 30% improvement in performace to its serial counterpart. The analyzes of the pathways produced by our simulation illustrates that thermodynamically folding and unfolding transitions are in good agreement with the real proteins. The protein model and the parallelized program developed in this dissertation will help us to understand the long time-scale behaviour of proteins, protein-protein interactions, protein structure predictions, as well as the change in stability of protein mutations.