A variety of model methods are proposed to analyze the high dimensional data.They provides new technical support for the study of gene mapping and complex disease traits inheritance,etc.In order to realize the localization of associated locus in genes,a mapping model was first constructed to recode the base pairs of each locus.Then,the method of combining quality control model with correlation analysis model is proposed to determine the correlation degree of locus.After that,the most relevant pathogenic locus associated with this genetic disease is screened out,according to the importance ranking of random forests.Finally,a high dimensional RBF neural network was designed to obtain the correlation coefficient of each locus and to explore the loci related to multiple traits of the disease.Combined with a variety of testing methods,the proposed model can accurately locate the loci and genes related to the disease.All kinds of models have strong generalization and are widely applicable to select large sample data with their own weights.%提出多种模型方法,对高维位点数据进行分析,为基因定位和复杂疾病性状遗传等方面的研究提供新的技术支持.为了实现关联位点在基因中的定位,首先建立映射模型,对每个位点的碱基对重新编码;然后,提出将质量控制模型与关联分析模型相结合的方法,确定位点的关联程度;随后利用基于随机森林的重要性排序,筛选与该遗传疾病最相关的致病位点;最后,设计出高维RBF神经网络,得到每个位点对性状的相关性系数,探索出与疾病多类性状相关的位点.结合多种检验方式,验证所建模型能够较为准确地定位与疾病相关的位点及基因.各类模型具有极强的推广性,广泛适用于筛选占有各自权值的大样本数据.
展开▼
