基于禁忌搜索的多因子降维在上位作用检测中的应用

基于禁忌搜索的多因子降维在上位作用检测中的应用 Tabu Search-Based Multifactor Dimensionality Reduction for Detecting Epistasis 杨俊 YANG Jun 1 first-author 殷建平 YIN Jianping 1 詹宇斌 ZHAN Yubin 1 国防科技大学计算机学院 国防科技大学计算机学院 College of Computer,National University of Defense Technology College of Computer,National University of Defense Technology 为了使多因子降维能够应用在大规模数据集的分析上,针对多因子降维穷举搜索单核苷酸多态（SNP）的组合的过程进行了改进,利用禁忌搜索的方法来代替穷举搜索.禁忌搜索以SNP的组合作为解,采用低阶的最优SNP组合作为高阶的初始解的一部分,随机替换当前解中的某个SNP作为邻域解,并采用全局多样化搜索,该方法在保证与多因子降维相当的分类正确率的基础上,显著地缩短了检测交互作用的时间. Instead of exhaustive search in multifactor dimensionality reduction we proposed tabu search for the combinations of single nucleotide polymorphism（SNP） for large-scale analysis.Tabu search uses combinations of SNP for solutions.The best low order solution is included in higher order initial solution.Neighborhood solutions are generated by randomly replacing a SNP in the current solution with a different SNP.We also used a diversification strategy.The proposed method has comparable power with multifactor dimensionality reduction（MDR） and it significantly reduce the computation time for detecting epistasis. 禁忌搜索 多因子降维 上位作用 特征选择 tabu search multifactor dimensionality reduction epistasis feature selection 10.14188/j.1671-8836.2011.06.003 R440 国家自然科学基金资助项目（61170287） 2011-06-01 2021-04-01 6 57