176
B. F. Voight
7.4.5 Integration of System-Based and “Omics” Approaches
with Genetics
Given the number of established associations of human traits that have been
discovered, along with the many polygenic contributions from numerous locations
in the genome which have not yet met genome-wide significance, a key line of
research is to utilize this collection of genetic data to uncover biological modules,
pathways, and processes which contribute to disease susceptibility. Computational
methods utilizing databases of text, gene expression, or protein-protein interaction
networks have been applied to some success in evaluating the hypothesis that genes
localizing nearby GWAS associations are more likely connected than expected
by chance (Zhu et al. 2008; Raychaudhuri et al. 2009; Rossin et al. 2011). The
results of applying these methods certainly suggest biological modules underlying
disease, but do not always return a clear picture of the nature of those pathways
(and causal candidates for them). Statistical methods such as gene-set enrichment
analysis offer a complementary approach to tests on prespecified networks for
the enrichment of statistical association (Segrè et al. 2010). This strategy offers a
principled approach to hypothesis testing but assumes that the biological networks
are known and can be specified ahead of time. Further research which incorporates
functional genomics data (expression from RNA sequencing, transcription factor
binding, histone occupancy, hypersensitivity assays) to further identify genes of
action, molecular mechanism, tissues of importance, and pathophysiology will be
increasingly available as large-scale experiments are underway (e.g., the genotypetissue expression project http://www.genome.gov/gtex/), the ENCODE project,
etc.). Given emerging evidence that suggests variants identified by GWAS are
enriched for expression quantitative trait loci (Nicolae et al. 2010), integrating
this information is almost certain to be of value. Recent success stories for fetal
hemoglobin levels (Bauer et al. 2013) and type 2 diabetes (Pasquali et al. 2014)
offer some promising early examples, whereby human genetic association data was
identified within annotated elements from high-throughput assays (in both cases,
enhancer elements), which were followed up and supported by additional functional
studies.
7.4.6 Uniting Findings from GWAS with Sequencing Studies
While GWAS studies have and will continue, they will gradually be met with highpowered and detailed high-throughput sequencing efforts for disease traits. The
primary aim of these studies is to comprehensively catalog variation at low and
rare allele frequencies and to test that spectrum of variation for its relationship to
disease. We should expect high power to discover variation implicated in diseases
but dramatically low power to demonstrate compelling association with complex
traits beyond a reasonable doubt (Guey et al. 2011). As a result, it will take some
time before sequencing studies come into their own. However, the discovery of
B. F. Voight
7.4.5 Integration of System-Based and “Omics” Approaches
with Genetics
Given the number of established associations of human traits that have been
discovered, along with the many polygenic contributions from numerous locations
in the genome which have not yet met genome-wide significance, a key line of
research is to utilize this collection of genetic data to uncover biological modules,
pathways, and processes which contribute to disease susceptibility. Computational
methods utilizing databases of text, gene expression, or protein-protein interaction
networks have been applied to some success in evaluating the hypothesis that genes
localizing nearby GWAS associations are more likely connected than expected
by chance (Zhu et al. 2008; Raychaudhuri et al. 2009; Rossin et al. 2011). The
results of applying these methods certainly suggest biological modules underlying
disease, but do not always return a clear picture of the nature of those pathways
(and causal candidates for them). Statistical methods such as gene-set enrichment
analysis offer a complementary approach to tests on prespecified networks for
the enrichment of statistical association (Segrè et al. 2010). This strategy offers a
principled approach to hypothesis testing but assumes that the biological networks
are known and can be specified ahead of time. Further research which incorporates
functional genomics data (expression from RNA sequencing, transcription factor
binding, histone occupancy, hypersensitivity assays) to further identify genes of
action, molecular mechanism, tissues of importance, and pathophysiology will be
increasingly available as large-scale experiments are underway (e.g., the genotypetissue expression project http://www.genome.gov/gtex/), the ENCODE project,
etc.). Given emerging evidence that suggests variants identified by GWAS are
enriched for expression quantitative trait loci (Nicolae et al. 2010), integrating
this information is almost certain to be of value. Recent success stories for fetal
hemoglobin levels (Bauer et al. 2013) and type 2 diabetes (Pasquali et al. 2014)
offer some promising early examples, whereby human genetic association data was
identified within annotated elements from high-throughput assays (in both cases,
enhancer elements), which were followed up and supported by additional functional
studies.
7.4.6 Uniting Findings from GWAS with Sequencing Studies
While GWAS studies have and will continue, they will gradually be met with highpowered and detailed high-throughput sequencing efforts for disease traits. The
primary aim of these studies is to comprehensively catalog variation at low and
rare allele frequencies and to test that spectrum of variation for its relationship to
disease. We should expect high power to discover variation implicated in diseases
but dramatically low power to demonstrate compelling association with complex
traits beyond a reasonable doubt (Guey et al. 2011). As a result, it will take some
time before sequencing studies come into their own. However, the discovery of