Abstract 20891: An Integrative Database for Robust Annotation of Findings in Genome-wide Studies of CVD and Subclinical Risk Factors
Background: Genome-wide association studies (GWAS) of CVD evaluate hundreds of thousands of SNPs and almost exclusively rely on single-SNP p-values to select “significant” ones, frequently missing biologically important ones.
Methods: A novel system was developed to provide Robust Annotation by Integrative Databases tailored to CVD genetics (RAIDCVD). Including ∼4.1 M SNPs, it identifies promising variables using information beyond single-variable p-values. In addition to standard annotations, RAIDCVD provides: 1) Imputed Functional Scores (IFS) estimated from cross-species conservation and transcription factor binding capacity; 2) Marker Ambiguity Scores (MAS) discriminating fine LD structure in different populations; 3) enrichment in pathways, networks, and interactions found in existing CVD studies. Weighted ranking combines information to prioritize SNPs. As an example, findings were annotated in a GWAS study of HTN in 3132 African Americans (AA) and 9315 Caucasians (CA).
Results: Briefly, 1) RAIDCVD provides data (eg, IFS) of functional potential for SNPs on current platforms (Figure); 2) For the example GWAS, at p<=10-3 we had 933 significant SNPs in CA and 1575 in AA, but only 2 in common; however with RAIDCVD annotation, 26 genes were shared (38% previously indicated in HTN); 3) IFS annotation improved many SNP ranks in known CVD genes (eg, rank of rs3777406 in IGF2R from 158 to 36 in CA); and 4) enrichment was detected in many gene-sets, including a novel network including membrane-associated guanylate kinase (DLG2) and calcium-dependent protein binding (ANXA6); 5) weighted ranking reduced the genes shared by races to 3 including PDGFD (highly expressed in heart) and A2BP1 (interacting with ATNX2 implicated in obesity).
Conclusion: RAIDCVD provides a novel tool for multi-source annotation in GWAS of CVD to select candidate SNPs for focused studies or to prioritize high-throughput data, leading to more robust findings across races.
- © 2010 by American Heart Association, Inc.