|
Activity Number:
|
357
|
|
Type:
|
Invited
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #308087 |
|
Title:
|
Multilevel Approaches to Genotype-Phenotype Association: VUS Classification and SNP Blocking
|
|
Author(s):
|
Giovanni Parmigiani*+ and Edwin Iversen and Xi Zhou and Daniele Fallin and Ingo Ruczinski and Cristian Pattaro
|
|
Companies:
|
Johns Hopkins University and Duke University and Cornell University and Johns Hopkins University and Johns Hopkins University and Eurac
|
|
Address:
|
Department of Pathology, Oncology and Biostatistics, Baltimore, MD, 21205,
|
|
Keywords:
|
variants of unknown significance ; bayesian classification ; hierarchical models ; SNPs
|
|
Abstract:
|
In this lecture I will present two applications of multilevel modeling in the area of genetic association studies. The first deals with determining whether the so-called "variants of unknown significance" on known disease genes are themselves disease-related. The second deals with the common case-control design for genome-wide association studies using SNPs. The common theme of the two applications is to use multilevel models to create, at a "lower" level data summaries that capture salient feature of complex data structures, such as families or structured regions in the genome. These parameters are in turn used as discriminating variables in a classification problem for which the lower level groups are the units.
|
