Novel methods to identify and replicate complex interactions in common disease

? DESCRIPTION (provided by applicant): Recent technological advances enable genotyping over 1-million human single-nucleotide polymorphisms (SNPs) on thousands of samples, and the characterization of an individual's entire three billion base pair genome is in sight. We are hindered in exploiting these laboratory advances to their fullest potential because strategies for analyzing these data have not kept pace, thus slowing progress toward improved understanding of the genetic contribution to common human disease. Additionally, the greater portion of heritability in complex traits has not been explained. Methodological approaches that embrace gene-gene (GxG) and gene- environment (GxE) interactions are essential. With the abundance of data available, we are in critical need of sophisticated analytical approaches that embrace this complexity. The current state-of-the-art methodologies do not allow for the modeling of complex interactions in diverse data types. Traditional GxG and GxE are performed using genetic coding methods (additive, dominant, recessive) that rely on overly simplistic assumptions about biology that may not be reflected in the data and are not flexible enough to allow for genetic heterogeneity. Two innovations provide potential solutions to these problems: (1) a novel weighted encoding method that allows for flexible genetic coding that is data-driven and (2) integration of interaction capability in BioBin, a biologically relevant variant-collapsing software. These novel approaches will be tested with three aims. First, current common genotype encoding methodologies will be evaluated and compared to our novel data-driven method for detecting SNP-SNP and SNP-environment interactions across comprehensive combination of simulated genetic models. Second, interaction capability will be incorporated into our BioBin software (BioBin-Interact) for rare and common variants and compared to traditional gene-based methods for detecting and replicating GxG and GxE, allowing for heterogeneity. Finally, the utility of the data-driven genetic encoding and BioBin-Interact will be tested across natural datasets for discovery and replication of GxG and GxE interactions predictive of age-related cataract, type 2 diabetes, and hypothyroidism. The long-term goal for these novel methods is to uncover complex interactions underlying common disease and ultimately assess individual disease-risk, contributing to a shift in the medical field toward personalized, preventie care.