SU109 - SCHIZOPHRENIA AND ADULT HEIGHT SHOW AN INVERSE POLYGENIC CORRELATION WITHIN SPECIFIC FUNCTIONAL DOMAINS OF THE GENOME

Abstract

Background Many epidemiological studies have reported on an inverse relationship between Schizophrenia (SCZ) and adult human height (H), two highly heritable traits. The largest investigation, a study of 1.35 million Swedish males, found that tall subjects (> 182 cm) had an approximately 15% reduction in risk of developing schizophrenia compared with short subjects ( Methods We applied GNOVA (GeNetic cOVariance Analyzer), a novel statistical framework that estimates annotation-stratified genetic covariance using GWAS summary statistics (Q. Lu et al., Biorxiv 2017). GNOVA models the genetic covariance between two traits as a linear system of equations using GWAS test-statistics and linkage disequilibrium (LD) across annotations. Annotation-stratified analyses can identify genetic overlap that would otherwise be missed across the whole genome. We used the most recent published GWAS summary statistics for both SCZ and H and estimated LD from 1000 Genome reference panel. We analyzed their genetic covariance (rho) stratified by Minor Allele Frequency (MAF) bins, the predicted functional genome, and transcriptomic and epigenomic annotations from a broad range of tissues and cell types from the GTEx and Roadmap Epigenomics consortia. Results Across the whole genome, we identified a significant inverse genetic relationship (rho= -0.0036, r= -0.03, P= 0.03), which was not detectable using the LD score regression method (LDSC; r= 0.00, P= 0.91). We found that the covariance is concentrated in variants of lower allele frequencies (MAF bin1: 5–18%, rho= -0.0044, P=0.003) that are predicted to be functional (rho= -0.0043, P=0.001). Across 53 tissues of GTEx, using only the subset of lower frequency variants, we observe the SCZ and H covariance in 13 tissues (P Discussion In summary, we are the first to report a significant inverse genetic relationship between SCZ and H. We show that this genetic overlap is small but significant and concentrated in variants of lower allele frequencies and specific functional domains of the genome. Current work involves Mendelian randomization analysis in a large SCZ cohort with anthropometric data available to further dissect their pleiotropic relationship. Our work confirms observations of decades of epidemiological studies and provides an integrative framework to investigate the shared genetic architecture between any two complex traits.