Background and Aims: Non-alcoholic fatty liver disease (NAFLD) is a progressive liver disease that ranges from simple steatosis to inflammation, fibrosis, and cirrhosis. To address the unmet need for new NAFLD biomarkers, we aimed to identify candidate biomarkers using publicly available RNA sequencing (RNA-seq) and proteomics data. Methods: A novel approach involving unsupervised gene clustering was performed using homogeneously processed and integrated RNA-seq data of 625 liver specimens to screen for NAFLD biomarkers, in combination with public proteomics data from healthy controls and NAFLD patients. Additionally, we validated the results in the NAFLD and healthy cohorts using enzyme-linked immunosorbent assay (ELISA) of plasma and immunohistochemical staining (IHC) of liver samples. Results: We generated a database (https://dreamapp.biomed.au.dk/NAFLD/) for exploring gene expression changes along NAFLD progression to facilitate the identification of genes and pathways involved in the disease's progression. Through cross-analysis of the gene and protein clusters, we identified 38 genes as potential biomarkers for NAFLD severity. Up-regulation of Quiescin sulfhydryl oxidase 1 (QSOX1) and down-regulation of Interleukin-1 receptor accessory protein (IL1RAP) were associated with increasing NAFLD severity in RNA-seq and proteomics data. Particularly, the QSOX1/IL1RAP ratio in plasma demonstrated effectiveness in diagnosing NAFLD, with an area under the receiver operating characteristic (AUROC) of up to 0.95 as quantified by proteomics profiling, and an AUROC of 0.82 with ELISA. Conclusions: We discovered a significant association between the levels of QSOX1/IL1RAP and NAFLD severity. Furthermore, the QSOX1/IL1RAP ratio shows promise as a non-invasive biomarker for diagnosing NAFLD and assessing its severity. (ChiCTR2300073940).