Objective To identify immune-related genes critical to asthma pathogenesis and construct a clinically applicable diagnostic model based on immune-gene signatures. Methods We first intersected 1639 immune-related genes (IRGs) with differentially expressed genes (DEGs) from the discovery dataset (GSE43696, n=128) to obtain differentially expressed IRGs (DE-IRGs). Expression stability was confirmed in the independent validation dataset GSE64913 (n=62). Least absolute shrinkage and selection operator (LASSO) regression and support-vector-machine (SVM) recursive feature elimination were applied to rank genes, followed by overlap with key modules identified by weighted gene co-expression network analysis (WGCNA). A logistic-regression diagnostic model was constructed using the optimal gene set, and its functional landscape was interrogated by gene-set enrichment analysis (GSEA). Finally, the model and the selected genes were cross-validated in ten transcriptomic cohorts encompassing eight distinct asthma phenotypes (total n=1137) and in lung tissue from an ovalbumin (OVA)-induced murine asthma model. Results A total of 38 DE-IRGs were identified. Among them, cholecystokinin (CCK), cellular retinol binding protein 2 (CRABP2) and C-X-C motif chemokine ligand 6 (CXCL6) were involved in immune-related processes and signaling pathways, which were of great significance in the diagnosis of asthma. The logistic regression diagnostic model based on three genes has shown good universality in various asthma samples. These three genes have also been verified to a certain extent in the lung tissues of OVA mice. Conclusion Integrative bioinformatics and in vivo validation establish CCK, CRABP2, and CXCL6 as a compact, biologically grounded immune-gene signature for asthma diagnosis and mechanistic investigation.
