Pathway Based Disease Analysis

• Current models for risk prediction use simple additive or multiplicative models for combining gene association information, hence they cannot account for relevant interactions between genes or pathways.
• Recent studies on pathway-centric modeling of genetic variations show that while particular genes affected very across patients, a set of core biological pathways is altered in some way in most or all patients.
• Hence, there is a need to identify critical biological nodes in pathways for the design of therapeutics.

Our approach is illustrated below:

• Our approach develops predictive network models of gene interactions formalized as Bayesian networks trained on tumor studies that contain genotypes of cancerous tissue.
• Our ultimate goal is a state-of-the-art risk model for multigenic diseases that effectively captures disease susceptibility.

The resulting models can be used to guide diagnosis and treatment as well as extend current knowledge of these diseases.