基于统计检验与机器学习的全基因组关联研究框架

发布时间：2018-01-10 10:33

本文关键词：基于统计检验与机器学习的全基因组关联研究框架　出处：《数学的实践与认识》2017年14期 　论文类型：期刊论文

【摘要】：采用统计检验和机器学习的方法来研究SNP或基因与疾病(可测性状)的关联性.先对SNP选择合适的数值编码方式,并设计了相应的统计检验流程,随后通过P值初步筛选出了与疾病或性状相关联的位点.在此基础上,对筛选出的位点,采用随机森林,XGBoost等机器学习方法,从样本外预测的角度判断SNP与疾病或性状的关联度.相关结果,显示发现运用该分析框架能较好地筛选出与疾病或性状关联的SNP(基因).并且框架由于考虑了多种分类模型,有着稳健性高,计算开销较小以及可以交叉比对等优势.框架未来在还可在金融,社交网络等方面发挥作用.
[Abstract]:Statistical test and machine learning were used to study the association of SNP or gene with disease (measurable traits). Firstly, the appropriate numerical coding method was selected for SNP, and the corresponding statistical test procedure was designed. Then the loci associated with disease or traits were preliminarily screened by P value. On this basis, the machine learning methods such as random forest XGBoost were used for the selected loci. The correlation degree between SNP and disease or trait was judged from the angle of prediction outside the sample. It was found that the SNPs associated with disease or traits could be well screened by using the framework, and the framework had high robustness because of the consideration of various classification models. The framework can also play a role in finance, social networks and so on in the future.
【作者单位】：上海大学经济学院;上海大学理学院;
【分类号】：O213;R3416;TP181
【正文快照】： 1 问题背景大量研究表明,人体的许多表型性状差异以及对药物和b;病的易感性等都可能与某些碱基位点相关联,或和包含有多个位点的基因相关联.因此,定位与性状或疾病相关联的位点在染色体或基因中的位置,能帮助研究人员了解性状和一些疾病的遗传机理,也能使人们对致病位点加以

【相似文献】