SUMMIT: An integrative approach for better transcriptomic data imputation improves causal gene identification

Genes with moderate to low expression heritability may explain a large proportion of complex trait heritability, but these genes are insufficiently captured in transcriptome-wide association studies (TWAS) partly due to the relatively small available reference datasets for developing expression genetic prediction models to capture the moderate to low genetically regulated components of gene expression. Here, we introduce a new method, Summary-level Unified Method for Modeling Integrated Transcriptome (SUMMIT), to improve the expression prediction model accuracy and the power of TWAS by using a large expression quantitative trait loci (eQTL) summary-level dataset. We applied SUMMIT to the eQTL summary-level data provided by the eQTLGen consortium, which involve 31,684 blood samples from 37 cohorts. Through simulation studies and analyses of GWAS summary statistics for 24 complex traits, we show that SUMMIT substantially improves the accuracy of expression prediction in blood, successfully builds expression prediction models for genes with low expression heritability, and achieves higher statistical power than several benchmark methods. In the end, we conducted a case study of COVID-19 severity with SUMMIT and identified 11 likely causal genes associated with COVID-19 severity.