Health Research and Policy


DATE: October 4, 2012
TIME: 1:15 - 3:00 pm
LOCATION: Li Ka Shing Center for Learning (LKSC)
291 Campus Dr, Room LK 209
1:15 pm - 3:00 pm
TITLE: An Overview of Developments in Statistical Genetics
SPEAKER: Laura Lazzeroni, Associate Professor (Biostatistics)
Department of Psychiatry and Behavioral Sciences, Stanford

  1. Extreme p-values: Recently, high throughput genomic data has shifted attention away from the traditional 0.05 significance level to very small p-values, on the order of 10^-8. I will discuss some surprising consequences of this change, including explicit results for the very large sampling variability of p-values that may help to explain the non-replication of many initial genetic findings.
  2. Generalized models for twin studies. Twin studies are often analyzed using standard software for mixed effects models. In real studies, data may not follow all assumptions of the model and computational difficulties may arise. Defries-Fulker regression is a computationally simple alternative, but does not formally incorporate covariates. I will discuss issues related to the parameter space and constraints of the twin model and propose a generalization of Defries-Fulker regression that addresses some of the problems encountered by mixed effects models.
  3. Interpretable models for gene-gene interactions. Several researchers have made recent proposals for gene-based, rather than SNP-based, genetic association tests. One of these is a principal component strategy proposed by Gauderman. I will use a pharmacogenetic study of antidepressant response to illustrate the method, showing how it can provide a straightforward and interpretable model for gene-gene interactions.
This work involves a number of collaborators including Ilana Belitskaya-Levy, Ying Lu, Amrita Ray, Martin Angst, David Clark, Greer Murphy, and Alan Schatzberg.

Back to Workshop in Biostatistics -->

Stanford Medicine Resources:

Footer Links: