Integrative analysis of proteome-wide association and transcriptome-wide association study identifies candidate brain proteins associated with insomnia

Abstract

Great progress has been made in identifying risk loci for insomnia by genome-wide association studies (GWAS) analysis, but its association with human brain proteome is unclear. Two insomnia GWAS summary datasets were derived from deCODE (n = 113,006) and 23andMe (n = 1,331,010). Two human brain proteomic datasets were obtained from ROS/MAP and Banner, and two reference datasets were obtained from brain RNA-seq (CBR) and RNA-seq splicing (CBRS). Proteome-wide association study (PWAS) was first used to detect brain proteins associated with insomnia at the translation level. Transcriptome-wide association study (TWAS) was then used to verify the results at the DNA level. Finally, brain imaging GWAS was used to explore the brain functional areas related to the identified brain proteins and genes in insomnia. PWAS identified 4 and 1 common proteins shared by two human brain proteomic datasets in insomnia GWAS dataset 1 and 2, such as ME1 (PDataset1−Banner−full=1.08×10−2, PDataset1−ROS/MAP−full=8.21×10−3). Further TWAS identified 5 and 1 candidate genes shared by the two reference expression profiles in dataset 1 and 2, like ICA1L (PDataset2−CBR=3.01×10−2, PDataset2−CBRS=3.24×10−2). CAMLG was observed to associate with insomnia in both PWAS (PDataset2−ROS/MAP=2.94×10−2) and TWAS (PDataset2−CBR=1.11×10−2). Comparing the results of PWAS and TWAS, there are 9 common proteins and genes shared by both two datasets, such as INPP4A. Brain imaging analysis found that insomnia associated proteins and genes were functionally related to cortex. Our results may reinforce the understanding of the etiology and pathophysiology of insomnia and provide promising brain protein targets for further therapeutic and mechanistic studies.