Health Research and Policy

Abstract

DATE: January 30, 2014
TIME: 1:15 - 3:00 pm
LOCATION: Medical School Office Building, Rm x303
TITLE: Inference of biological networks using graphical models
SPEAKER: Christine Peterson, Postdoctoral Fellow
Center for Computational, Evolutionary and Human Genomics,
Dept of Health Research and Policy, Stanford

Graphical models are a class of statistical models which describe the conditional dependence relationships among a set of random variables. Technological innovations in the biological sciences have allowed us to measure the levels of many genes, proteins, or metabolites for each subject. In this context, understanding how the levels of these molecules are related through biological networks is of critical importance in understanding the biological mechanisms of disease and treatment. Since these networks are complex systems, they can be difficult to infer given a limited number of observations. In this workshop, I will discuss both standard approaches for the inference of graphical models and my own research in this field. Specifically, we will discuss methods I developed in the Bayesian framework that allow incorporation of prior information on network relationships and problem structure to improve inference given limited sample sizes. I will illustrate these methods with applications to the inference of cellular metabolic and protein-protein networks.

Suggested readings:
Friedman, J., Hastie, T. and Tibshirani, R. (2007). Sparse inverse covariance estimation with the graphical lasso. Biostatistics. 9 (3), p.432-441.

Peterson, C., Vannucci, M., Karakas, C., Choi, W., Ma, L. and Maletic-Savatic, M. (2013). Inferring metabolic networks using the Bayesian adaptive graphical lasso with informative priors. Statistics and Its Interface. 6 (4): 547-558.

Stanford Medicine Resources:

Footer Links: