Integrating Proteomics and GWAS to Identify Key Tissues and Genes Underlying Human Complex Diseases

• Published in: Biology (Basel, Switzerland) Vol. 14; no. 5; p. 554
• Main Authors: Xue, Chao, Zhou, Miao
• Format: Journal Article
• Language: English
• Published: Switzerland MDPI AG 16.05.2025; MDPI
• Subjects: Arthritis; associated genes; Bipolar disorder; Cardiovascular disease; complex diseases; Coronary artery disease; Coronary heart disease; Crohn's disease; Diabetes mellitus (non-insulin dependent); Disease susceptibility; disease-associated tissues; fine-mapping; Gene expression; Gene mapping; Genes; Genetic aspects; Genetic research; Genome-wide association studies; genome-wide association study; Heart diseases; Mental disorders; Molecular modelling; Peptide mapping; Protein expression; Proteins; Proteomics; Rheumatoid arthritis; Rheumatoid factor; Ribonucleic acid; RNA; Schizophrenia; Standard scores; Tissues; Transcriptomics; Type 2 diabetes; disease-associated tissues; proteomics; complex diseases; associated genes; genome-wide association study; fine-mapping
• ISSN: 2079-7737; 2079-7737
• DOI: 10.3390/biology14050554

Background: The tissues of origin and molecular mechanisms underlying human complex diseases remain incompletely understood. Previous studies have leveraged transcriptomic data to interpret genome-wide association studies (GWASs) for identifying disease-relevant tissues and fine-mapping causal genes. However, according to the central dogma, proteins more directly reflect cellular molecular activities than RNA. Therefore, in this study, we integrated proteomic data with GWAS to identify disease-associated tissues and genes. Methods: We compiled proteomic and paired transcriptomic data for 12,229 genes across 32 human tissues from the GTEx project. Using three tissue inference approaches—S-LDSC, MAGMA, and DESE—we analyzed GWAS data for six representative complex diseases (bipolar disorder, schizophrenia, coronary artery disease, Crohn’s disease, rheumatoid arthritis, and type 2 diabetes), with an average sample size of 260 K. We systematically compared disease-associated tissues and genes identified using proteomic versus transcriptomic data. Results: Tissue-specific protein abundance showed a moderate correlation with RNA expression (mean correlation coefficient = 0.46, 95% CI: 0.42–0.49). Proteomic data accurately identified disease-relevant tissues, such as the association between brain regions and schizophrenia and between coronary arteries and coronary artery disease. Compared to GWAS-based gene association estimates alone, incorporating proteomic data significantly improved gene association detection (AUC difference test, p = 0.0028). Furthermore, proteomic data revealed unique disease-associated genes that were not identified using transcriptomic data, such as the association between bipolar disorder and CREB1. Conclusions: Integrating proteomic data enables accurate identification of disease-associated tissues and provides irreplaceable advantages in fine-mapping genes for complex diseases.