Genome-wide association studies (GWAS) and next-generation sequencing are now commonplace despite a lack of comprehensive bioinformatics approaches for relating genotype to phenotype. The common method of analysis is to employ parametric statistics and then adjust for the large number of tests performed to limit false-positives. This agnostic approach is preferred by some because no assumptions are made about which genes or genomic regions might be important. The goal of our proposed research program continuation is to develop and evaluate a bioinformatics approach that analyzes genetic associations in the context of expert knowledge about biochemical pathways, gene function and experimental results using gene set enrichment (GSE) methods. An important challenge for success in this domain is the quality of the expert knowledge that is available in public databases such as Gene Ontology (GO). We first propose to develop and evaluate a novel Data-driven Ontology Refinement Algorithm (DORA) for improving the quality of genetic and genomic annotations (AIM 1). Improving the quality of annotations will in turn improve GSE results. We will then develop a comprehensive bioinformatics approach to the analysis of high-throughput genetic association results that considers functional DNA elements, genes, and gene function as important contexts. We will first determine whether considering data from the Encyclopedia of DNA Elements (ENCODE) database improves GSE analysis at the level of gene regions (AIM 2). Next we will determine whether using GO annotations refined by our novel DORA algorithm (DORA-GO) improves GSE analysis at the gene set level above and beyond that provided by GO (AIM 3). We will determine the validity of these methods by assessing the replication of the results in independent data (AIM 4).
AIMS 1 -4 will be accomplished using several large population-based genetic studies of pre-clinical cardiovascular disease (CVD) as measured by left ventricular mass (LVM). Our working hypothesis is that we will obtain more replicated and hence more real genetic associations using our novel bioinformatics methods that embrace, rather than ignore, prior biological knowledge.
The bioinformatics methods and software developed and distributed as part of this research project will play an important role in advancing our ability to fully exploit genome-wide association data for common, complex diseases.
Showing the most recent 10 out of 157 publications
