Identifying Nonparallel Differentially Methylated DNA Regions Using Smoothing Spline ANOVA Model

作者:sds_admin 发布时间:2017-01-06

Title: Identifying Nonparallel Differentially Methylated DNA Regions Using Smoothing Spline ANOVA Model
Speaker: Xin Xing( Department of Statistics at the University of Georgia)
Time: 2016.12.12 14:00-15:00
Address: Room 102, Zibin Building, No. 220 Handan Rd, Shanghai
Abstract: DNA methylation can repress certain genes’ expression and is thus a key factor in regulating gene expression.  Aberrant DNA methylation can cause many human diseases including cancer. Identifying differentially methylated regions (DMRs) is thus crucial for finding possible therapeutic targets. We develop a nonparametric composite hypothesis test for identifying DMRs. There are two major difficulties in the non-parametric inference problem. The first one is the distribution of the testing statistic is hard to obtain. Moreover, it is computational cost is extremely high accessing the distribution by re-sampling methods like bootstrapping and permutation test. The second difficulty is that the test is usually lack of power. To overcome the difficulties, we derived the asymptotic distribution of the test statistics and characterize the power of test using the mini-max separation rate. Empirically, the proposed test is applied on whole genome bisulphite sequencing data to study the aberrant DNA methylation pattern in chronic lymphocytic leukemia (CLL) patients.