Integrated analysis of scRNA-seq and bulk RNA-seq identifies matrisome-related biomarkers for prognostic stratification and immune landscape in lung adenocarcinoma

limma, with prognostic genes identified using univariate Cox. A LASSO-multivariate Cox model was built and validated using Kaplan-Meier, receiver operating characteristic, and external datasets. Functional enrichment (GO/KEGG/GSEA), immune infiltration (ssGSEA/ESTIMATE/CIBERSORT), tumor mutational burden, immunotherapy response, and drug sensitivity were assessed. Consensus clustering defined molecular subtypes. Single-cell analysis (Seurat/SCISSOR) identified risk-associated cells, with AUCell and CellChat evaluating activity and cell communication. A 7-gene risk model (ANGPTL4, C1QTNF6, CCL20, CLEC3B, FCN1, LAMA3, PRELP) stratified LUAD patients into distinct survival groups (area under the curve > 0.7). Low-risk patients showed higher immune infiltration, lower TIDE scores (suggesting better immunotherapy response), and reduced sensitivity to Axitinib/Gefitinib. Single-cell analysis implicated fibroblasts and myeloid cells in high-risk profiles, with activated MIF/TGF-β pathways. This integrated transcriptomic and single-cell model predicts LUAD prognosis and immune landscape, guiding personalized therapy.