Research On Gene Association Analysis Based On Ultra-high Dimensional Uncertain Data

Master student: Lele Ma Supervisor: XiaozhenJiang;WenjunXiongMajor: Applied Statistics Research Direction: Biostatistics Grade: 2018With the advancement of modern biological technology,more and more people pay attention to the association between the genes carried by individuals and specific diseases in the process of studying genetic diseases,and find the disease-causing genes of individuals through the method of genome-wide association analysis Site,this method is also widely used effectively in reality.The increasing size of data has brought challenges to data analysis.When doing genome-wide association analysis,when the dimensions are low,many scholars will consider the three possibilities of genetic models: recessive models,additive models,and dominant models.When performing genetic testing on ultra-high-dimensional genetic data,Analysis needs to take into account the uncertainty of the genetic model.In addition,when doing genetic association analysis,there are cases where the genotype is also uncertain,so further research on the association analysis of ultra-high-dimensional gene data when the genotype is uncertain is needed.This paper compares the simulation of variable screening with the SIS method under and without consideration of the genetic model.The results show that more pathogenic gene loci can be selected under the consideration of the genetic model.When the genetic model is uncertain,continue to use the LASSO and SCAD methods for variable selection.Both methods have a high degree of recognition for the locus gene model;when the genotype is uncertain,the expected value of the genotype probability value is used to represent the gene Type data,and then use deterministic independent screening and LASSO and SCAD methods for variable selection.The results show that the variable selection method has a good effect on the screening of pathogenic gene loci.