The complementary roles of rare variant burden scores and common variant polygenic risk scores in genetic risk prediction of complex disorders

Abstract

Polygenic risk scores (PRS) derived from common variants have improved risk prediction for complex disorders, however they capture only a fraction of heritability. One of the reasons for this is the exclusion of rare variants, despite growing evidence of their substantial contribution to genetic liability. Here we implemented the Independent Outlier Gene Count (IOGC) - a score summarising the individual burden of rare variants associated with outlying gene expression - and measured the amount of liability explained by this rare-variant score, alone and in combination with common-variant PRS, for schizophrenia, major depressive disorder, and hypertrophic cardiomyopathy in the UK Biobank. IOGC and PRS were minimally correlated (|r|<0.2), indicating the two scores capture complementary information. For schizophrenia, a model including both IOGC and PRS explained 1.13 times the adjusted liability than PRS alone. For hypertrophic cardiomyopathy, a 1.48-fold increase was observed for the combined model. People with high IOGC showed a 64% and 46% higher positive predictive value for schizophrenia and HCM, respectively, as compared to those with low IOGC. No improvement was observed for depression. These findings demonstrate that integrating rare- and common-variant scores enhances genetic risk prediction for complex conditions with a strong genetic component, supporting the inclusion of rare variant burden metrics in future genomic risk models.