170
B. F. Voight
regulation by microRNAs MIR-137 for schizophrenia (Williams et al. 2011) as well
as Lin28/let-7 for glucose metabolism and risk to diabetes (Zhu et al. 2011). In
addition to recapitulating known and discovery of unknown biology, these studies
have also recapitulated association at established pharmacological interventions for
disease (statins, thiazolidinediones, sulfonylureas, and several others), despite the
relatively weak effect sizes associated at each locus. Also, genetic studies on adverse
drug response (SEARCH Collaborative Group et al. 2008) or treatment response
(Ge et al. 2009) have identified factors strongly predictive and relevant in clinical
outcomes. Interestingly, genes nearest to established associations may not always
be causal: a recent report for an association for T2D indicated strong evidence for
long-range enhancer activity regulating a gene further away (IRX3/5) than the most
proximal one (FTO). This interaction and subsequently follow-up effort implicated
brown fat and the biology of non-shivering thermogenesis as a result (Smemo et
al. 2014). As further results from GWAS are taken forward into model systems
(Musunuru et al. 2010) or in vitro studies to further elucidate the mechanism, one
can expect the pace of understanding to proceed at a steady pace given the labor and
time these experiments require.
7.3.6 The Importance of Phenotypic Collection in Study Designs
Another clear but obvious message is that phenotypic ascertainment can greatly
impact the genetic study and associations discovered. An initial example of the
importance was first observed in the first published GWAS studies for type 2
diabetes (T2D) (The Diabetes Genetics Initiative 2007). In that study, the design
explicitly matched the body mass index (BMI) of cases with that of controls, with
the goal of identification of variants for T2D not mediated through obesity. As a
result, the study directly controlled for the effect of a strong genetic effect for
obesity, variation near the FTO/IRX3/5 locus. Subsequently, other studies without
phenotypic ascertainment in this way discovered and replicated variation at that site,
both with obesity and T2D (Frayling et al. 2007). As the field progresses toward
next-generation genetic studies, with more sophisticated phenotypic measurements
or extreme sampling (Guey et al. 2011), careful attention to the collection process
(and under what conditions those collections occur or are associated with genotype)
will be increasingly important. In addition, collection of a battery of additional
trait measurements (endophenotypes) will be valuable for diseases where clinical
boundaries are hard to describe or assign with precision (e.g., psychiatric conditions,
Crohn’s disease vs. ulcerative colitis)
7.3.7 Challenges with Interpreting the Distribution of Effects
and Frequencies
The increasing number of polymorphisms associated with complex traits facilitates
the first glimpses of the distribution of effect sizes and the frequency spectrum
B. F. Voight
regulation by microRNAs MIR-137 for schizophrenia (Williams et al. 2011) as well
as Lin28/let-7 for glucose metabolism and risk to diabetes (Zhu et al. 2011). In
addition to recapitulating known and discovery of unknown biology, these studies
have also recapitulated association at established pharmacological interventions for
disease (statins, thiazolidinediones, sulfonylureas, and several others), despite the
relatively weak effect sizes associated at each locus. Also, genetic studies on adverse
drug response (SEARCH Collaborative Group et al. 2008) or treatment response
(Ge et al. 2009) have identified factors strongly predictive and relevant in clinical
outcomes. Interestingly, genes nearest to established associations may not always
be causal: a recent report for an association for T2D indicated strong evidence for
long-range enhancer activity regulating a gene further away (IRX3/5) than the most
proximal one (FTO). This interaction and subsequently follow-up effort implicated
brown fat and the biology of non-shivering thermogenesis as a result (Smemo et
al. 2014). As further results from GWAS are taken forward into model systems
(Musunuru et al. 2010) or in vitro studies to further elucidate the mechanism, one
can expect the pace of understanding to proceed at a steady pace given the labor and
time these experiments require.
7.3.6 The Importance of Phenotypic Collection in Study Designs
Another clear but obvious message is that phenotypic ascertainment can greatly
impact the genetic study and associations discovered. An initial example of the
importance was first observed in the first published GWAS studies for type 2
diabetes (T2D) (The Diabetes Genetics Initiative 2007). In that study, the design
explicitly matched the body mass index (BMI) of cases with that of controls, with
the goal of identification of variants for T2D not mediated through obesity. As a
result, the study directly controlled for the effect of a strong genetic effect for
obesity, variation near the FTO/IRX3/5 locus. Subsequently, other studies without
phenotypic ascertainment in this way discovered and replicated variation at that site,
both with obesity and T2D (Frayling et al. 2007). As the field progresses toward
next-generation genetic studies, with more sophisticated phenotypic measurements
or extreme sampling (Guey et al. 2011), careful attention to the collection process
(and under what conditions those collections occur or are associated with genotype)
will be increasingly important. In addition, collection of a battery of additional
trait measurements (endophenotypes) will be valuable for diseases where clinical
boundaries are hard to describe or assign with precision (e.g., psychiatric conditions,
Crohn’s disease vs. ulcerative colitis)
7.3.7 Challenges with Interpreting the Distribution of Effects
and Frequencies
The increasing number of polymorphisms associated with complex traits facilitates
the first glimpses of the distribution of effect sizes and the frequency spectrum