Meta-analysis indicates that common variants at the DISC1 locus are not associated with schizophrenia

Abstract

Several polymorphisms in the Disrupted-in-Schizophrenia-1 (DISC1) gene are reported to be associated with schizophrenia. However, to date, there has been little effort to evaluate the evidence for association systematically. We carried out an imputation-driven meta-analysis, the most comprehensive to date, using data collected from 10 candidate gene studies and three genome-wide association studies containing a total of 11 626 cases and 15 237 controls. We tested 1241 single-nucleotide polymorphisms in total, and estimated that our power to detect an effect from a variant with minor allele frequency >5% was 99% for an odds ratio of 1.5 and 51% for an odds ratio of 1.1. We find no evidence that common variants at the DISC1 locus are associated with schizophrenia.

Figure 1

Reported Disrupted-in-Schizophrenia-1 (DISC1) associations including both family and association studies from populations of all ethnicities and indicating regions reported as being significantly associated with schizophrenia.^{, , , , , , , , , , , , ,} These includes family-based linkage studies and studies in non-European populations, which were not included in this study. The numbers to the right of the graph indicate numbers of cases/controls in the study.

Figure 2

Estimated power as a function of minor allele frequency at the causal allele, showing results for different odds ratios. The power was simulated using variants and haplotypes from the 1000 Genomes data and the curves have been fitted as smooth splines of degree 6.

Figure 3

Pooled P-values for association. Each point represents a single-nucleotide polymorphism (SNP). The horizontal axis indicates chromosomal position, and the vertical axis indicates negative log₁₀ of the fixed-effect pooled P-value for that SNP. The black points show the imputed SNPs and the gray points show the subset of the imputed SNPs, which were genotyped in at least one study. The horizontal dashed lines show the corrected significance level, which is equivalent to a 5% gene-wide significance level for each data set.

Figure 4

Quantile–quantile (QQ) plot of P-values. QQ plot of the P-values shown in Figure 2. The solid line shows the expected values under the null hypothesis of no association and the dashed lines show the upper and lower 95% quantiles.

Figure 5

Wakefield approximate Bayes factors for each single-nucleotide polymorphism (SNP). Calculated from the random effects pooled P-values and Z-scores assuming a normal prior on the log-OR, with mean 0 and variance of 0.015. A Bayes factor >0 is evidence for association, whereas a Bayes factor <0 is evidence against.