An improved simulated annealing algorithm and its application to minimizing the energy of protein tertiary structure

This paper provides an improved Simulated Annealing (SA) algorithm --- Parallel Simulated Annealing Adaptive Neighborhood /Records Crossover (PSAAN/RX). In this algorithm there are several processes of SA working in parallel. During the processes, record operator and arithmetic crossover operator are performed. In addition, a neighborhood control rule is used. Numerical tests confirm the effectiveness of all these improvements. The proposed algorithm is then applied to energy minimization of protein energy function. Target proteins are Metenkephalin and C-peptide. Comparing PSAAN/RX to another two algorithms, it is found that PSAAN/RX can achieve conformations with lower energy values and PSAAN/RX has higher success rates. This algorithm can be used as an ad-initio method in small protein structure prediction.