基于改进遗传聚类的生物基因多序列分类方法  

A Methodology for Modeling Multiple Sequence Classification of Biological Genes Based on Improved Genetic Clustering

作　　者：钟娉婷 徐胜超 ZHONG Pingting;XU Shengchao(Institute of Artificial Intelligence,Guangzhou Huashang College,Guangzhou 5511300)

机构地区：[1]广州华商学院人工智能学院,广州511300

出　　处：《计算机与数字工程》2024年第12期3512-3516,3522,共6页Computer & Digital Engineering

基　　金：国家自然科学基金面上项目(编号:61972444)资助。

摘　　要：研究了一种基于改进遗传聚类的生物基因多序列分类方法。使用隐马尔可夫模型对每个基因序列进行建模;通过量子优化算法改进了遗传算法的选择、交叉和变异机制,用于搜索生物基因多序列分类模型的群中心。通过动态调整量子旋转角,引入自适应变异算子,以增强对最优群中心的全局搜索能力。在改进的遗传算法对K-means聚类算法的群中心进行优化后,实现多基因序列的精确分类。实验结果表明,该研究方法能够实现生物基因多序列分类,该分类内样本分布密集,分类间样本具有高度区分性,平均F1得分指标达到93.45%。A multi-sequence classification model of biological genes based on improved genetic clustering is investigated to provide basic data for genetic engineering.A hidden Markov model(HMM)is used to model each gene sequence.The mechanism of selection,crossover and variation of the genetic algorithm is improved by using the quantum optimization algorithm.The improved genetic algorithm is used to search the cluster center.By dynamically adjusting the quantum rotation angle,the adaptive mutation operator is introduced to enhance the global search ability of the optimal cluster center.After the improved genetic algorithm is used to optimize the cluster center of the K-means clustering algorithm.The accurate classification of multiple gene sequences is achieved.The experimental results show that the method can achieve biological gene multi sequence classification,the samples with-in the class are densely distributed,and the samples between classes have a high degree of discrimination,with the average F1 score index reaching93.45%.

关 键 词：基因序列 隐马尔可夫模型 状态转移矩阵 K-MEANS聚类算法 群中心 量子遗传算法 

分 类 号：TP393.04[自动化与计算机技术—计算机应用技术]