|
Activity Number:
|
471
|
|
Type:
|
Invited
|
|
Date/Time:
|
Wednesday, August 3, 2016 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Mental Health Statistics Section
|
|
Abstract #318254
|
View Presentation
|
|
Title:
|
Covariates Modulated False Discovery Rate in Hidden Markov Random Field Model
|
|
Author(s):
|
Richard Levine and Wesley Thompson and Carrie Bearden and Dorothy Parker and Rong Zablocki*
|
|
Companies:
|
San Diego State University and University of California at San Diego and University of California at Los Angeles and University of California at San Diego and University of California at San Diego
|
|
Keywords:
|
false discovery rate ;
Hidden Markov Random Field ;
Bayesian mixture model ;
Gene-network
|
|
Abstract:
|
Large-scale hypothesis testing is facing the challenge of properly controlling for type I error. Traditional multiple-comparison procedures tend to be underpowered. Procedures that control false discovery rate (fdr) are more powerful; yet most of proposed methods treat all hypothesis tests as exchangeable and independent, ignoring auxiliary covariates that may influence the distribution of the test statistics and ignore the prior knowledge of the network structure. The current work incorporates Hidden Markov Random Field (HMRF) into a Bayesian two-group mixture model for data with network structure. The simulations are carried out to compare the performance in terms of sensitivity, specificity and false discovery proportion (FDP) between the new method and existing methods. It is also demonstrated in a real data application in gene-network for identifying disease-associated genes.
|
Authors who are presenting talks have a * after their name.
