Abstract
Structural and microstructural variations of human brain are heritable and highly polygenic traits, with hundreds of associated genes founded in recent genome-wide association studies (GWAS). Using gene expression data, transcriptome-wide association studies (TWAS) can prioritize these GWAS findings and also identify novel gene-trait associations. Here we performed TWAS analysis of 211 structural neuroimaging phenotypes in a discovery-validation analysis of six datasets. Using a cross-tissue approach, TWAS discovered 204 associated genes (86 new) exceeding Bonferroni significance threshold of 1.37*10−8 (adjusted for testing multiple phenotypes) in the UK Biobank (UKB) cohort, and validated 18 TWAS or previous GWAS-detected genes. The TWAS-significant genes of brain structures had been linked to a wide range of complex traits in different domains. Additional TWAS analysis of 11 cognitive and mental health traits detected 69 overlapping significant genes with brain structures, further characterizing the genetic overlaps among these brain-related traits. Through TWAS gene-based polygenic risk scores (PRS) prediction, we found that TWAS PRS gained substantial power in association analysis compared to conventional variant-based PRS, and up to 6.97% of phenotypic variance (p-value=7.56*10−31) in testing datasets can be explained by UKB TWAS-derived PRS. In conclusion, our study illustrates that TWAS can be a powerful supplement to traditional GWAS in imaging genetics studies for gene discovery-validation, genetic co-architecture analysis, and polygenic risk prediction.
Footnotes
↵7 These authors jointly supervised this work: Yun Li, Hongtu Zhu.
List of Alzheimer’s Disease Neuroimaging Initiative (ADNI) and Pediatric Imaging, Neurocognition and Genetics (PING) authors provided in the supplemental materials.
