|
Activity Number:
|
188
- Bayesian Application to Biological and Health Sciences
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 4, 2020 : 10:00 AM to 2:00 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
|
Abstract #313145
|
|
|
Title:
|
Identifying Individual Rare Variants with a Fast Bayesian Variable Selection Method for Trios: Using TRIO_RVEMVS
|
|
Author(s):
|
Duo Yu* and Kusha Mohammadi and Margaret Steiner and Abigail C Sedory and Michael Swartz and Matthew D. Koslovsky
|
|
Companies:
|
University of Texas Health Science Center at Houston and Univeristy of Texas- Health Science Center at Houston and The George Washington University and University of Texas Health Science Center at Houston and University of Texas Health Science Center At Houston, School of Public Health and The Rice University
|
|
Keywords:
|
Rare variants;
EMVS;
Trio family;
Spike and slab priors;
Bayesian variable selection
|
|
Abstract:
|
Trio Rare Variants Expectation-Maximization Variable Selection (TRIO_RVEMVS) is a novel approach that jointly models common and rare variants (RVs) to analyze trio data. Contrary to traditional genetic association methods for RVs, TRIO_RVEMVS uses spike and slab priors to quickly identify both genetic regions with unexpected risk burden from RVs and the individual RVs associated with the disease. Using Cosi2 to simulate realistic trio families (n = 350), we compare our method to PEDGENE by computing a weighted average correct association probability (WACAP). TRIO_RVEMVS and PEDGENE had a WACAP of 66.37% and 55.73% when analyzing gene regions using both common and rare variants, respectively. When analyzing RVs only, the WACAP of TRIO_RVEMVS and PEDGENE was 52.07% and 52.65%, respectively. The average true positive rate for individual RVs using TRIO_RVEMVS was 4.33% with an average false positive rate of 0.13%. Thus, TRIO_RVEMVS compares well with one of the top methods for identifying associated regions with RV burden while also capturing associated RVs. We applied our method to a trio dataset from the Gabriella Miller Kids First Pediatric Research Project.
|
Authors who are presenting talks have a * after their name.
