174
B. F. Voight
assumption that pathway modules and mechanisms that contribute genetic risk
to multiple traits are the same across traits (or at least, for related traits). It
is still a theoretical possibility that, locus-to-locus, different associations point
to different (but proximally located) genes in an interval. Multiple SNPs could
still potentially involve a single gene actor, but implicate different mechanisms
important for different traits. For example, the ~1 Mb region upstream of the
well-known proto-oncogene, c-MYC, has multiple associations to different cancers
across tissues (Ghoussaini et al. 2008), with a leading hypothesis that this is due to
cis-expression modules for various regulatory elements which are tissue-specific. A
primary analysis will be to evaluate the pathways around regions where genomewide significance for one trait has been established and multiple associations with
additional related traits also reside.
An additional line of inquiry will involve comparisons across trait groups
where prior evidence of biological overlap across traits will certainly be of
interest—for example, autoimmune associations with metabolic and psychiatric
disease or cancer with metabolic and cardiovascular traits. There are at least
some examples in data where potentially coincident associations have been found
(e.g., the regions around BCL11A and the CDKAL1 with Crohn’s disease and
T2D; the MHC region with multiple autoimmune disease and psychiatric disease). While the exact gene candidates and mechanisms remain to be elucidated, exploratory analysis at this stage can begin to rule out (or hone in on)
regions of the genome that could contribute risk, collectively. By combining
data from multiple genetic scans for psychiatric disease (schizophrenia, bipolar
disorder, attention deficit disorder, autism), new genetic loci directly contributing
to the shared burden of disease have been reported (Williams et al. 2011), with
additional support for a shared, common genetic comorbidity among several
pairs of traits (Cross-Disorder Group of the Psychiatric Genomics Consortium
2013).
7.4.3 Genetic Fine Mapping for Complex Trait Loci
One activity that can be expected uniformly across complex traits will be to
apply fine-mapping procedures to characterize the allele spectrum, identify new
variants, and generate hypotheses of causal variants and genes at established loci.
Early and highly illustrative examples of the strategy have already been performed
(Graham et al. 2007), and the first large-scale fine-mapping studies amassing data
across multiple ethnicities are starting to be reported (DIAGRAM Consortium
2014). The strategy begins by first sequencing a large panel of individuals at
association regions to discover all polymorphic sites that could be related to
disease, and second genotype all variants in a large number of samples and
perform systematic conditional analyses to identify signals which explain the lead
association signal, as well as additional associations independent of the leading
signal. This specific design has been prohibitive until very recently, because of
two key developments: first, availability of custom array genotyping technologies to
B. F. Voight
assumption that pathway modules and mechanisms that contribute genetic risk
to multiple traits are the same across traits (or at least, for related traits). It
is still a theoretical possibility that, locus-to-locus, different associations point
to different (but proximally located) genes in an interval. Multiple SNPs could
still potentially involve a single gene actor, but implicate different mechanisms
important for different traits. For example, the ~1 Mb region upstream of the
well-known proto-oncogene, c-MYC, has multiple associations to different cancers
across tissues (Ghoussaini et al. 2008), with a leading hypothesis that this is due to
cis-expression modules for various regulatory elements which are tissue-specific. A
primary analysis will be to evaluate the pathways around regions where genomewide significance for one trait has been established and multiple associations with
additional related traits also reside.
An additional line of inquiry will involve comparisons across trait groups
where prior evidence of biological overlap across traits will certainly be of
interest—for example, autoimmune associations with metabolic and psychiatric
disease or cancer with metabolic and cardiovascular traits. There are at least
some examples in data where potentially coincident associations have been found
(e.g., the regions around BCL11A and the CDKAL1 with Crohn’s disease and
T2D; the MHC region with multiple autoimmune disease and psychiatric disease). While the exact gene candidates and mechanisms remain to be elucidated, exploratory analysis at this stage can begin to rule out (or hone in on)
regions of the genome that could contribute risk, collectively. By combining
data from multiple genetic scans for psychiatric disease (schizophrenia, bipolar
disorder, attention deficit disorder, autism), new genetic loci directly contributing
to the shared burden of disease have been reported (Williams et al. 2011), with
additional support for a shared, common genetic comorbidity among several
pairs of traits (Cross-Disorder Group of the Psychiatric Genomics Consortium
2013).
7.4.3 Genetic Fine Mapping for Complex Trait Loci
One activity that can be expected uniformly across complex traits will be to
apply fine-mapping procedures to characterize the allele spectrum, identify new
variants, and generate hypotheses of causal variants and genes at established loci.
Early and highly illustrative examples of the strategy have already been performed
(Graham et al. 2007), and the first large-scale fine-mapping studies amassing data
across multiple ethnicities are starting to be reported (DIAGRAM Consortium
2014). The strategy begins by first sequencing a large panel of individuals at
association regions to discover all polymorphic sites that could be related to
disease, and second genotype all variants in a large number of samples and
perform systematic conditional analyses to identify signals which explain the lead
association signal, as well as additional associations independent of the leading
signal. This specific design has been prohibitive until very recently, because of
two key developments: first, availability of custom array genotyping technologies to