相似度检测法在遗传性疾病和遗传位点上的应用

Similarity detection method in the determination of disease-causing gene and genetic loci

在生物学中,寻找致病位点、表现型性状差异及疾病的易感性等都属于分类问题,目的是把等待处理的样本进行分类,找出异常数据。相似度的优点在于类与类之间寻找异常数据,将相似度检测运用于致病位点的查询,结合卡方检验和支持向量机,能更精确地找到致病位点。实验中相似度检测得到的致病位点和卡方检验得到的致病位点都有可能是该病的致病位点,致病位点可能存在于两者的集合之中,同时根据支持向量机的小样本、非线性问题中表现出许多特有的优势,在实验中使用支持向量机进行建模,在模型精度较高的基础上确定各个位点的权重,权重大的位点对结果的影响明显,达到筛选出最有可能致病的位点的目的。

In biology,searching for the pathogenic loci,the difference of phenotype and the susceptibility of disease is a kind of classification problem,aiming to classify the samples and to find the abnormal data.The similarity can find the abnormal data between classes.Therefore,combining similarity detection method with chi square test and support vector machine can find the pathogenic loci more accurately.Both the pathogenic loci detected by similarity detection method and chi square test would be the disease loci of the disease,and the disease loci may exist in the collection of both.With the advantages in the small sample and nonlinear problems,support vector machine is used for the modeling.Based on the established model of high precision,we select the most likely loci of disease by determining the weights of loci because the effect of the loci with large weight on the results is significant.