associations” [11]. The replication samples should ideally be
larger to be well powered to identify spuriously associated
SNPs (seeNote 13).- The guidelines to conduct replication studies were suggested
by NCI-NHGRI Working Group on Replication in Association
Studies [12].
3.6 Statistical
Analysis
Larger sample sizes will have a greater possibility of identifying
genetic factors that have a more modest effect. The common way
is that, for instance, each group does their own GWA analysis, and
then the data from several studies is combined together by
performing a meta-analysis of the results for each genetic variant.
To obtain a statistical significant result, we will show a typical
procedure for GWAS.- Create the Manhattan plot by HaploView v4.1 [13] or the
quantile-quantile (Q-Q) plot generated by R 2.8.1 for data
visualization, which can provide a visual summary of associa-
tion test results that draws immediate attention to any signifi-
cant regions.
Q-Q plot: One way to evaluate whether there is an excess of
significant results at a given threshold is to plot theP-values
that result from the test of association against theP-values from
a uniform distribution. The correction for population stratifi-
cation can reduce the excess result and excess associations that
are false positive and that are not due to true genetic signals. - Perform fine mapping around the newly identified susceptibil-
ity gene locus by genotyping tag SNPs and performing impu-
tation [14](seeNote 14). - Meta-analysis is useful for the replication of initial association
results, and can increase power and opportunity to identify
novel signals associated with a disease. When performing
meta-analysis, one has to concern about heterogeneity between
the studies. For example, when the WTCCC performed a
GWAS of T2D, they showed strong evidence of association of
variants at the FTO locus. However, a couple of other studies
that were doing association analysis of T2D at the same time
did show the same result. It is because the WTCCC were more
obese than the controls in that study, whereas in the other
diabetes studies, their case-control selection had been more
balanced with respect to body size. Identifying this source of
heterogeneity between the studies led to the identification of
this BMI gene.
(a) Meta-analysis can be conducted by PLINK to combine
multiple data from GWA studies and provide a quantita-
tive evaluation of the consistency/inconsistency or
heterogeneity of the results across multiple datasets (see
Note 15).
104 Michelle Chang et al.