中文摘要：

针对当前检测基因-基因交互作用方法中存在的一些缺陷,提出一种基于群智能和冲突规避策略的基因-基因交互作用检测方法（DEIBSC）.以SNP（single nucleotide polymorphism）为研究对象,从大量SNP中选出具有显著基因-基因交互作用的SNP组.初始化多个SNP组作为初值,同时产生多条搜索路径,利用得分单调递增原则寻找问题的解,通过冲突规避策略和群智能动态调整搜索路径的方向,使得到的解更能反映基因-基因交互作用在基因组范围内分布的情况.在仿真数据和真实数据上的实验证实,本文方法在统计能力上可以和SNPHarvester方法相比,在效率上有明显优势,得到的结果能够更广泛地代表基因-基因交互作用在基因组的分布.

英文摘要：

In order to overcome the deficiencies of present methods for detecting gene-gene interactions,we propose a new method（DEIBSC） based on swarm intellegence and collision avoidance strategy.Focusing on SNPs,our method selects those SNP groups having significant gene-gene interactions from a mass of SNPs.It initializes a certain amount of SNP groups,and then generates multiple search paths.Using score monotone increasing principle and adjusting search paths dynamically,our method makes the solutions reflect the distribution of gene-gene interactions in genome.Experimentations on real data and simulated data testify that our method is comparable to SNPHarvester in statistic power and that it has significant advantage in efficiency,from which the solutions are able to reflect the distribution of gene-gene interactions in genome.