Concepedia

Bivariate causal mixture model quantifies polygenic overlap between complex traits beyond genetic correlation

Oleksandr Frei, Dominic Holland, Olav B. Smeland, Alexey Shadrin, Chun Chieh Fan, Steffen Maeland, Kevin S. O’Connell, Yunpeng Wang, Srdjan Djurovic, Wesley K. Thompson,

Nature Communications · 2019 · 443 citations · 47 references

DOI Full text

Open access

TL;DR

Accumulating evidence from genome‑wide association studies suggests a vast amount of shared genetic influence among complex human traits and disorders, such as mental illnesses. The study introduces MiXeR, a statistical tool that quantifies polygenic overlap independent of genetic correlation, aiming to enhance understanding of cross‑trait genetic architectures. MiXeR uses GWAS summary statistics to quantify polygenic overlap and presents the results as a Venn diagram of unique and shared components across traits. MiXeR estimates that schizophrenia and bipolar disorder share 6.2 K of their 8.3 K and 6.4 K causal variants, and that schizophrenia and educational attainment share 8.3 K causal variants despite near‑zero genetic correlation, with 2.5 K variants unique to education.

Abstract

Abstract Accumulating evidence from genome wide association studies (GWAS) suggests an abundance of shared genetic influences among complex human traits and disorders, such as mental disorders. Here we introduce a statistical tool, MiXeR, which quantifies polygenic overlap irrespective of genetic correlation, using GWAS summary statistics. MiXeR results are presented as a Venn diagram of unique and shared polygenic components across traits. At 90% of SNP-heritability explained for each phenotype, MiXeR estimates that 8.3 K variants causally influence schizophrenia and 6.4 K influence bipolar disorder. Among these variants, 6.2 K are shared between the disorders, which have a high genetic correlation. Further, MiXeR uncovers polygenic overlap between schizophrenia and educational attainment. Despite a genetic correlation close to zero, the phenotypes share 8.3 K causal variants, while 2.5 K additional variants influence only educational attainment. By considering the polygenicity, discoverability and heritability of complex phenotypes, MiXeR analysis may improve our understanding of cross-trait genetic architectures.