172
B. F. Voight
blood pressure measures, triglyceride levels with high-density lipoprotein (HDL)
cholesterol, body mass index and waist-to-hip or waist circumference measures,
and many other examples.
Observed epidemiological correlation between traits can be intuitively thought
of as the aggregated effect of all genetic (and environmental) perturbations across
traits. However, even if the aggregated effect results in a significant correlation
between traits, individual genetic factors are not required to contribute to both and
could even have opposing effects. An outstanding example of this phenomenon
comes again from one of the first published well-powered GWAS (The Diabetes
Genetics Initiative 2007). That study observed a coding polymorphism at the
glucokinase regulatory protein, GCKR, initially associated with triglyceride levels
but subsequently with fasting glucose levels (Orho-Melander et al. 2008) and type
2 diabetes susceptibility (Dupuis et al. 2010). However, in contrast to observational
epidemiology predicting a positive correlation among triglycerides, fasting glucose
levels, and susceptibility to T2D, the allele associated with increased triglycerides
was associated with decreased fasting glucose levels and lower risk to T2D at
genome-wide levels of significance. A resolution to this paradox was proposed
after careful mutational and mechanistic study indicating that the effect is due
to a reduction in regulatory effect by fructose-6 phosphate-mediated inhibition of
GCKR, resulting in increased glucokinase activity in the liver (Beer et al. 2009).
This effect is predicted to enhance glycolytic flux, promoting hepatic glucose
metabolism and elevated concentrations of malonyl-CoA, a substrate for de novo
lipogenesis (Beer et al. 2009). While the story around GCKR is an unusual one,
other examples are emerging (Kilpeläinen et al. 2011). It is this spectrum of unique
variation discovered by GWAS, along with the notion that such variation empirically
exists and can be found, that will contribute to the understanding of the mechanisms
underlying disease susceptibility and trait biology.
7.4
What Lies Just Beyond the Horizon for GWAS?
In the immediate term, there are several genetic experiments that follow from what
has already been learned from GWAS and should be anticipated in the coming
years. The central theme that underlies each of them is the strategy to collect (and
genetically test) large genetic data sets within previously uncharacterized human
populations, join them together with other genetic data sets across populations
and phenotypes, and finally integrate them with new, high-throughput genomics
technologies. Systematic application of each of these threads is aimed to (1) help
identify the polygenic contribution of diseases by systematic genetic study, (2)
pinpoint causal genetic variation and hypotheses of the mechanism by fine mapping
established loci for disease, and (3) generate hypotheses for networks and pathways
by integrating with functional data sets.
B. F. Voight
blood pressure measures, triglyceride levels with high-density lipoprotein (HDL)
cholesterol, body mass index and waist-to-hip or waist circumference measures,
and many other examples.
Observed epidemiological correlation between traits can be intuitively thought
of as the aggregated effect of all genetic (and environmental) perturbations across
traits. However, even if the aggregated effect results in a significant correlation
between traits, individual genetic factors are not required to contribute to both and
could even have opposing effects. An outstanding example of this phenomenon
comes again from one of the first published well-powered GWAS (The Diabetes
Genetics Initiative 2007). That study observed a coding polymorphism at the
glucokinase regulatory protein, GCKR, initially associated with triglyceride levels
but subsequently with fasting glucose levels (Orho-Melander et al. 2008) and type
2 diabetes susceptibility (Dupuis et al. 2010). However, in contrast to observational
epidemiology predicting a positive correlation among triglycerides, fasting glucose
levels, and susceptibility to T2D, the allele associated with increased triglycerides
was associated with decreased fasting glucose levels and lower risk to T2D at
genome-wide levels of significance. A resolution to this paradox was proposed
after careful mutational and mechanistic study indicating that the effect is due
to a reduction in regulatory effect by fructose-6 phosphate-mediated inhibition of
GCKR, resulting in increased glucokinase activity in the liver (Beer et al. 2009).
This effect is predicted to enhance glycolytic flux, promoting hepatic glucose
metabolism and elevated concentrations of malonyl-CoA, a substrate for de novo
lipogenesis (Beer et al. 2009). While the story around GCKR is an unusual one,
other examples are emerging (Kilpeläinen et al. 2011). It is this spectrum of unique
variation discovered by GWAS, along with the notion that such variation empirically
exists and can be found, that will contribute to the understanding of the mechanisms
underlying disease susceptibility and trait biology.
7.4
What Lies Just Beyond the Horizon for GWAS?
In the immediate term, there are several genetic experiments that follow from what
has already been learned from GWAS and should be anticipated in the coming
years. The central theme that underlies each of them is the strategy to collect (and
genetically test) large genetic data sets within previously uncharacterized human
populations, join them together with other genetic data sets across populations
and phenotypes, and finally integrate them with new, high-throughput genomics
technologies. Systematic application of each of these threads is aimed to (1) help
identify the polygenic contribution of diseases by systematic genetic study, (2)
pinpoint causal genetic variation and hypotheses of the mechanism by fine mapping
established loci for disease, and (3) generate hypotheses for networks and pathways
by integrating with functional data sets.