基于群智能的蛋白质耐热温度预测研究

为直接利用序列和结构信息预测蛋白质耐热温度，提出了基于群智能的蛋白质耐热温度预测方法。基于多元线性回归模型，利用人工蜂群与粒子群混合算法，优化了蛋白质的耐热温度与氨基酸含量的多元线性回归模型的参数，得到蛋白质的耐热温度。此外，通过加入蛋白质的氨基酸网络拓扑属性，提高了蛋白质耐热温度的预测准确性。对耐温蛋白质，网络拓扑属性的加入使得蛋白质耐热温度的预测值偏差和真实值偏差之间的相关系数增加到0.71，平均预测率增加到0.88；耐热蛋白质的相关系数增加到0.75，平均预测率增加到0.91。氨基酸网络拓扑属性的引入为预测蛋白质耐热温度提供了新的视角。

Prediction research of protein melting temperature based on swarm intelligence algorithm

In order to predict the protein melting temperature directly from the term of the sequence and structure of the protein, the prediction of protein melting temperature based on swarm intelligence algorithm is proposed. Using the hybrid algorithm that combines artificial bee colony and particle swarm, which optimizes the parameters of multivariate linear regression model, the paper calculates the protein melting temperature based on amino acid content. Additionally, by adding amino acid network topological properties to amino acid content, the prediction accuracy is greatly improved. For mesophilic protein, the correlation coefficient between the predicted value deviation and the real value deviation increases to 0.71, the average of prediction accuracy increases to 0.88 compared with the result using amino acid content; and the correlation coefficient increases to 0.75, the average of prediction accuracy increases to 0.91 of thermophilic protein. Amino acid network topological properties offer an innovation perspective for the prediction of protein melting temperature.