Title: "Control Type I Error in Genome-wide Association Studies"
Speaker: Shizhong Xu; Department of Botany and Plant Sciences, UC Riverside, Riverside, CA

Place: SMITH (SMTH) Hall 108
Date: March 25, 2014; Tuesday
Time: 4:30pm


The mixed model approach to genome-wide association studies is very efficient in controlling the additive polygenic background effect by fitting a polygenic variance into the mixed model. The method cannot to control the polygenic background effect due to dominance and epistasis. It also fails to control false positives potentially caused by linkage disequilibrium. We proposed to incorporate dominance and epistasis into the polygenic background to further reduce the residual variance. We also proposed a moving window genome scanning method to reduce the Type I error without sacrifice of statistical power. We defined a window of fixed or variable length on the genome and estimate the genetic variance for a marker in the middle of the window and the genetic variances for the two markers bracketing the window. We only tested the significance of the marker in the middle of the window while treating the variances of the flanking markers as mechanisms to reduce false positives caused by linkage disequilibrium. The entire genome is scanned by the moving window to detect significant association of the marker in the middle of the window and a quantitative trait. Simulation studies showed that this method has significantly reduced genome-wide Type I error rate compared with the original mixed model methodology of genome-wide association studies.

Associated reading:
1. S. Xu. 2013.Mapping Quantitative Trait Loci by Controlling Polygenic Background Effects. Genetics. 195: 1209-1222.

Click here www.stat.purdue.edu/~doerge/BIOINFORM.D/SPRING14/sem.html for a full schedule of BIOINFORMATICS SEMINARS, past and present.