Additive Model and Variable Selection for Uncertain Genotypes
思敏 钟 · Statistics and Applications · 2021
全基因组关联分析(GWAS)是研究复杂疾病相关位点的有效方法.在基因不确定情形下，传统方法利用基因填补方式估计基因概率，继而展开后续基因关联分析。我们对大样本基因考虑一个非参数可加模型对可加分量维数大而非零加性分量数目小的基因数据进行建模，其中加性分量利用B样条基函数的线性组合工具来近似拟合基因概率对性状表征的效应关系；选择非零分量是利用组Lasso惩罚来获得初始估计量。最后我们利用蒙特卡洛模拟证明，可加模型的组lasso方法在基因表达样本中的效果良好。 Genome-wide association analysis (GWAS) is an effective method to study the associated loci of complex diseases. In the case of genetic uncertainty, the traditional method uses the gene filling method to estimate the gene probability, and then carries out the subsequent gene association analysis. We used a nonparametric additive model to model the data of large samples of genes with large additive component dimensions but small non-zero-additive component numbers. The additive component was used as a linear combination tool of B-spline basis function to approximate the effect relationship of gene probability on trait characterization. The group Lasso penalty was used to obtain the initial estimator for selecting the non-zero component. Finally, Monte Carlo simulation was used to demonstrate that the group Lasso method of the additive model performed well in gene expression samples.