This project seeks to contribute to the understanding of the genetic basis for schizophrenia (SCZ) and bipolar disease (BD). To achieve this goal, we attempt to amplify genetic signals of modest effect in an SCZ/BD cohort, by analyzing such a cohort together with a well powered study of another (relevant) phenotype. The strategy relies on clarifying etiologic pathways to illness by looking at overlaps with a genetically well characterized correlated trait - in this case height (H). Large scale epidemiological studies suggest that increased H is associated with a decreased risk of SCZ. Given that BD is comorbid with SCZ, H might share (as suggested by our pilot analyses) causal pathways with each of these two psychiatric disorders. The genetic meta-analysis of H is probably the largest to be published, which ensures that it has good power to detect even modest genetic signals. Thus, H is a good candidate for a phenotype to be tested for genetic overlap with SCZ/BD, where by genetic overlap we mean the SNP/genes which significantly affect both phenotypes, not just one. Critically, this overlap can allow us to clarify etiologic pathways to SCZ/BD that might be quite difficult to detect in other ways. Furthermore, we suggest that our method can yield at least two classes of etiologic pathways. We provide preliminary evidence that pathways where the genetic effects on the two phenotypes are concordant (i.e., in the same direction) are very different from pathways where they are discordant. Thus, we suggest that, to avoid pathway heterogeneity, it is advisable for the concordant and discordant signals to be analyzed separately in pathway analyses. To uncover a part of the genetic architecture of SCZ/BD we employ a two steps process using only publicly available univariate summaries from relevant meta-analyses. In the first step we evaluate i) the genetic overlap of each disease with a) H and b) between SCZ and BD. In the second step, the most promising concordant and discordant overlap signals are used in separate gene set analyses to uncover whether these signals are enriched in certain molecular pathways. To assess the overlap between SCZ/BD and H, we develop novel statistical methods to i) increase the genetic data resolution by imputing summary statistics at unobserved SNPs based only on the summary statistics at observed SNPs and ii) obtain the genetic overlap between multiple phenotypes which is not overly influenced by a strong signal coming from just one phenotype.
Results from meta-analytic studies of major psychiatric disorders and other phenotypes are coming online. Unfortunately, even these meta-analytic studies of psychiatric disorders might not have power to detect causal variants of modest effects. However, a careful analysis of meta-analytic studies of psychiatric disorders together with well powered studies of other relevant phenotypes has the potential to amplify some of the genetic pathways harboring those variants with modest effects.
|Edwards, Alexis C; Bacanu, Silviu-Alin; Bigdeli, Tim B et al. (2016) Evaluating the dopamine hypothesis of schizophrenia in a large-scale genome-wide association study. Schizophr Res 176:136-40|
|Lee, Donghyung; Bigdeli, T Bernard; Williamson, Vernell S et al. (2015) DISTMIX: direct imputation of summary statistics for unmeasured SNPs from mixed ethnicity cohorts. Bioinformatics 31:3099-104|
|Williamson, V S; Mamdani, M; McMichael, G O et al. (2015) Expression quantitative trait loci (eQTLs) in microRNA genes are enriched for schizophrenia and bipolar disorder association signals. Psychol Med 45:2557-69|
|Lee, Donghyung; Williamson, Vernell S; Bigdeli, T Bernard et al. (2015) JEPEG: a summary statistics based tool for gene-level joint testing of functional variants. Bioinformatics 31:1176-82|
|Bigdeli, T Bernard; Bacanu, Silviu-Alin; Webb, Bradley T et al. (2014) Molecular validation of the schizophrenia spectrum. Schizophr Bull 40:60-5|
|Bacanu, Silviu-Alin; Chen, Xianging; Kendler, Kenneth S (2013) The genetic overlap between schizophrenia and height. Schizophr Res 151:226-8|
|Lee, Donghyung; Bigdeli, T Bernard; Riley, Brien P et al. (2013) DIST: direct imputation of summary statistics for unmeasured SNPs. Bioinformatics 29:2925-7|