What will diabetes genomes tell us?

Abstract

A new generation of genetic studies of diabetes is underway. Following from initial genome-wide association (GWA) studies, more recent approaches have used genotyping arrays of more densely spaced markers, imputation of ungenotyped variants based on improved reference haplotype panels, and sequencing of protein-coding exomes and whole genomes. Experimental and statistical advances make possible the identification of novel variants and loci contributing to trait variation and disease risk. Integration of sequence variants with functional analysis is critical to interpreting the consequences of identified variants. We briefly review these methods and technologies and describe how they will continue to expand our understanding of the genetic risk factors and underlying biology of diabetes.

Figure 1

The top sequences represent the observed genotypes of two individuals from different studies that used different genome-wide arrays. Different markers are available on the arrays, and in this example, only one shared marker is available for the two individuals. The reference haplotypes contain many more markers (SNPs, and, more recently, insertions or deletions) than the genotyped samples. Reference haplotypes may be from the HapMap Project, the 1000 Genomes Project, or other sequenced or densely genotyped samples. Haplotypes from the reference samples that are consistent with the observed individual haplotypes are highlighted. These reference haplotypes are used to fill in (impute) the unobserved genotypes in the study individuals (bottom). More than one reference haplotype can be consistent with an individual’s phased genotypes. To account for this, imputation programs provide the probability of each genotype. Asterisks indicate markers that are in perfect linkage disequilibrium within the reference panel; an ‘A’ at the first asterisk marker is always observed with ‘G’ at the second asterisk marker and ‘G’ at the third asterisk marker.