Gene Association Analysis Of Complex High-dimensional Generalized Additive Model With Multi-dimensional Phenotype

Genome wide association analysis(GWAS)is one of the most important methods for the study of traits and genetic diseases.At present,most of the researches on genotypes are based on the hypothesis of regression analysis or genetic model.Gene loci are multidimensional,and phenotypic data are usually multidimensional.Multi dimensional phenotypic data in the relationship known model,the association solution is feasible,but in reality,the specific model of association is usually unknown,the model is assumed to be linear or other known models,which may affect the actual fitting effect.In high-throughput sequencing,sometimes we can't get all the specific genotypes,usually we can only get the genotype probability,so we need to fill in the genotype information.In this paper,we try to study these problems,establish the association model between gene and multi-dimensional phenotype by nonparametric method,and analyze the uncertainty of genotype.The research order of this paper is as follows: first,data of undetermined genotypes are embedded in multidimensional phenotypes;Secondly,the relationship model between gene and multi-dimensional phenotype was established by nonparametric model,and the multi-dimensional phenotype was correlated;Thirdly,the sparsity of high-dimensional gene loci after transformation is discussed,and some unimportant variables are screened out;Fourthly,the summary and Prospect of the full text are given,and the feasibility of the model in gene data is discussed.The simulation results show that the nonparametric model performs well in multi-dimensional phenotypic data,the associated loci are basically significant,and the accuracy is in line with expectations.