This paper reports the results of a genome-wide association study of seven common diseases—including bipolar disorder—carried out by the Wellcome Trust Case Control Consortium (WTCCC). Although it comes at the end of a recent wave (tsunami?) of papers reporting genome-wide association results, this important work deserves careful study. Aficionados will appreciate the rich methodological detail and the careful approach to many issues—population stratification, the relationship of SNPs to genes, correction for multiple testing—that worry those of us who do these studies. Everyone will get a sense of the power of this new approach to detect genetic signals that have long eluded the best efforts of the field.
Although it returned fewer highly significant “hits” than most of the other diseases, the bipolar disorder sample (ably collected and phenotyped by Craddock and colleagues) nevertheless yielded many interesting findings. These are worthy of follow-up either by virtue of their statistical significance, apparent biological plausibility, or concordance with other studies.
On the issue of concordance, at first glance there may not appear to be much overlap between the WTCCC findings and those of our own recent genome-wide association study of bipolar disorder (Baum et al., 2007). But first glances can be misleading. For example, some readers may already have noticed that one gene, DFNB31, that was placed near the top of the list in our study, is also implicated by one of the top 10 SNPs (by p-value) in the WTCCC results. DFNB31 encodes whirlin, a key component of the Usher Protein Complex that modulates Wnt signaling and plays an important role in synaptic development (Kremer et al., 2006). As we work our way through the differing platforms (we used the Illumina HumanHap 550K, whose SNPs overlap only ~20 percent with those on the Affymetrix chip used by the WTCCC), we will likely encounter additional points of agreement. The initial goal is to find SNPs that meet the very high standards of replication called for in a recent paper from the NCI-NHGRI Working Group on Replication in Association Studies (Chanock et al., 2007). But in a complex disease like bipolar disorder, support may come in more complex forms: multi-locus interactions, gene-wise replications, genes that overlap in common pathways, etc. While we must guard against confirmatory bias, we do not want to discard valuable evidence merely on the basis of current statistical opinion, however distinguished its source.
Those who want to compare the results themselves can access both our results and those of the WTCCC at the respective websites. The WTCCC site even includes a set of imputed genotype counts for many HapMap SNPs that will greatly facilitate cross-platform comparisons. More data is on the way from GAIN (Genetic Association Information Network), which has a strong data sharing policy, and from other groups, who will need to share their data or risk becoming irrelevant.
Whatever the outcome, with the advent of the WTCCC and the ongoing GAIN studies, psychiatric genetics has now unmistakably entered the fraternity of “big science.” The initiation rites have not been without pain, but the advantages of membership—at least in the short term—will be great. In the long term, we must also make sure that the small research laboratories and individual psychiatric investigators do not get cut out. Their insight and ingenuity will be needed if we are to succeed in quickly turning the results of big science to the benefit of patients.
View all comments by Francis McMahon