Differential Gene Expression Profiling and Machine Learning-Based Discovery of Key Genetic Markers in VTE and CKD

Hui LiCai LinJunJie Kuang^*

• First People's Hospital of Huizhou City, Huizhou, China

Venous thromboembolism (VTE) and chronic kidney disease (CKD) are complex disorders with multifaceted genetic underpinnings. In this study, we delved into gene expression patterns associated with these diseases by contrasting patients with control groups. Differential analysis pinpointed 637 differentially expressed genes (DEGs) between VTE patients and controls, of which 224 were downregulated and 413 upregulated. For CKD patients, 671 DEGs were observed, encompassing 572 downregulated and 99 upregulated genes. Functional enrichment analyses via GO and KEGG pathways unveiled that DEGs in VTE patients were predominantly enriched in cellular cytoplasmic translation, immune-responsive cell activation, and ribosomal subunits. KEGG analysis further illustrated VTE DEGs primarily involved in oxidative phosphorylation, carcinogenic reactive oxygen species, and the 2019 coronavirus disease signaling pathways. CKD DEGs, on the other hand, leaned towards the regulation of muscle contraction, cation-transporting ATPase complex, and endopeptidase inhibitor activity. They were also prevalent in pathways like Pancreatic secretion, Protein digestion and absorption, and Vascular smooth muscle contraction. Intersecting DEGs from CKD and VTE identified 23 overlapping genes, inclusive of CCNL2, HNRNPA0, PI4KA, FOS, HBD, among others. Their significant differential expression across the diseases was affirmed by Wilcox test. Subsequent application of machine learning algorithms, namely LASSO, SVM-RFE, and Random Forest, highlighted two pivotal genes: HNRNPA0 and PI4KA with unparalleled AUCs of 1.000 for both. A diagnostic nomogram for CKD, rooted in these feature genes, was developed, revealing high concordance between predicted and observed CKD risks. This was further cemented by ROC analysis. GSEA enrichment and immune cell infiltration investigations linked HNRNPA0 and PI4KA to diverse pathways and immune cell associations. Validation in datasets like GSE37171 and GSE48000 consistently reflected lower expressions of these genes in CKD samples versus controls. At a single-cell level, clustering of cells from both sets produced 11 unique clusters, annotated to eight primary cell types. The key genes' expression across these cells was also mapped. In sum, our findings illuminate crucial genes and their roles in VTE and CKD, offering a potent predictive model for CKD diagnosis based on key genetic markers.