基于SAPSO-LSSVM的蛋白质模型质量评估

Predicting protein model quality assessment by SAPSO-LSSVM

针对传统蛋白质模型质量评估没有考虑同源信息的问题,提出了一种基于LS-SVM评估蛋白质模型质量的方法。综合模拟退火(simulated annealing,SA)算法跳出局部最优解和粒子群(particle swarm optimization,PSO)算法收敛速度快的特点,提出了模拟退火粒子群(SAPSO)算法。利用SAPSO算法来优化LS-SVM参数C和γ,最后得到最优模型来评估蛋白质模型质量。实验结果表明,经SAPSO优化LS-SVM参数所得到的模型评估预测误差较小,且预测值更稳定。

As the traditional methods evaluated the quality of models without considering the source information,this paper designed a new algorithm to access the quality of a protein structure based on least squares support vector machines. Firstly,considering the characteristics that simulated annealing( SA) algorithm could jump out of the local optimal solution and that particle swarm optimization( PSO) algorithm had a fast convergence speed,it developed a simulated annealing particle swarm optimization( SAPSO) algorithm; the next it optimized the parameters of LS-SVM that included C,γ by SAPSO algorithm; lastly the algorithm built an optimal model,which used to predict the quality score of a protein structure model. The experimental results show that the model of SAPSO algorithm optimized the parameters has less error in predicting protein structures and its prediction is more stable.