Objective: To use genome-wide fixed marker arrays and improved analytical tools to detect genetic associations with type 2 diabetes in a carefully phenotyped human sample.
Research design and methods: A total of 1,087 Framingham Heart Study (FHS) family members were genotyped on the Affymetrix 100K single nucleotide polymorphism (SNP) array and examined for association with incident diabetes and six diabetes-related quantitative traits. Quality control filters yielded 66,543 SNPs for association testing. We used two complementary SNP selection strategies (a "lowest P value" strategy and a "multiple related trait" strategy) to prioritize 763 SNPs for replication. We genotyped a subset of 150 SNPs in a nonoverlapping sample of 1,465 FHS unrelated subjects and examined all 763 SNPs for in silico replication in three other 100K and one 500K genome-wide association (GWA) datasets.
Results: We replicated associations of 13 SNPs with one or more traits in the FHS unrelated sample (16 expected under the null); none of them showed convincing in silico replication in 100K scans. Seventy-eight SNPs were nominally associated with diabetes in one other 100K GWA scan, and two (rs2863389 and rs7935082) in more than one. Twenty-five SNPs showed promising associations with diabetes-related traits in 500K GWA data; one of them (rs952635) replicated in FHS. Five previously reported associations were confirmed in our initial dataset.
Conclusions: The FHS 100K GWA resource is useful for follow-up of genetic associations with diabetes-related quantitative traits. Discovery of new diabetes genes will require larger samples and a denser array combined with well-powered replication strategies.