改进型蚁群聚类算法在单核苷酸多态性(SNPs)数据分析中的应用  

Improved Ant Colony Clustering Algorithm in Single-nucleotide Polymorphisms(SNPs) Data Analysis Application

作　　者：姜龙训 张玲[1] 

机构地区：[1]首都医科大学公共卫生学院,北京市右安门外西头条10号100069 [2]北京市丰台区南苑社区卫生服务中心

出　　处：《中国数字医学》2015年第5期77-80,共4页China Digital Medicine

基　　金：国家自然科学基金资助项目-盐敏感性高血压候选基因的验证及其风险预测模型的构建(编号:81373076);北京市教育委员会科技计划面上项目-原发性高血压盐敏感性遗传多态性的分子流行病学研究(编号:SQKM201210025010)~~

摘　　要：目的:改进经典蚁群聚类算法(LF算法),应用到盐敏性高血压SNPs数据分析,为探讨高通量SNPs统计分析提供新思路。方法:改进LF算法,利用Mat 1ab8.0软件对改进后算法进行编程,对335个盐敏性高血压样本进行聚类分析,并通过潜在类别分析的结果进行比较。结果:成功改进LF算法并实现软件化界面。采用新算法将所有样本分成2个类别,第一类169份样本,第二类166份样本,与潜在类别分析法结果进行一致性检验,Kappa值为0.93,P<0.001,并通过两类人群SNPs概率分布差异统计学检验,筛选出3个SNPs:rs848307、rs1739843、rs1010069,明确其在分类中的重要作用。结论:蚁群聚类算法具有思维独特、计算自动化、易于改进等特点,在高通量SNPs数据分析及其他基因组学相关领域有广阔的应用前景。Objective： Improved classical ant colony clustering algorithm applied to salt-sensitive hypertension single nucleotide polymorphism data analysis, high-throughput SNPs investigate complex diseases statistical analysis provides a new approach to support. Methods Performed classical ant colony clustering algorithm optimized for improved Matlab8.0 software using the improved algorithm can be programmed for 335 salt-sensitive hypertension samples containing 29 SNPs cluster analysis data, and through potential category results of the analysis were compared and evaluated the improved algorithm.Results Improved ant colony algorithm to establish and implement the software interface. Using this algorithm, al the samples successful y into two categories, the first class of 169 samples （50.4%）, the second 166 samples （49.6%）. Using latent class analysis Cluster analysis results obtained for the first class 174 samples （51.9%）, the second 158 samples （47.1%）. Both methods consistency test, Kappa value 0.93, P〈0.001, and the distribution of the difference between the probability of significant SNPs tested by two populations screened in 29 SNPs rs848307, rs1739843, rs10100693 SNPs in its important role in the classification. Conclusion Improved ant colony clustering algorithm has unique thinking, computing automation, easy to improve other characteristics, form their own advantage, SNPs in high-throughput data analysis and other related fields of genomics has broad application prospects.

关 键 词：蚁群优化算法 单核苷酸多态性(SNPs) 聚类分析 

分 类 号：R197.3[医药卫生—卫生事业管理]