Integrative single cell and machine learning analysis identifies fibroblast centric hub genes and therapeutic candidates in lung adenocarcinoma and ischemic cardiomyopathy

Background Ischemic cardiomyopathy (ICM) and lung adenocarcinoma (LUAD) are distinct diseases but may share fibroblast mediated mechanisms involving tissue remodeling and inflammation. However, their common molecular features at single cell resolution remain insufficiently defined. Methods We integrated single cell RNA sequencing datasets from ICM and LUAD with bulk transcriptomic profiles from TCGA LUAD, GSE18842, and GSE48166. Differential expression analysis identified 36 shared core genes. A total of 127 combinations of machine learning algorithms and feature selection methods were benchmarked to identify six hub genes: KRTCAP3, RUNX2, DBF4B, C11orf80, COL3A1, and ZFP64. UCell, scMetabolism, SCISSOR, Monocle 3, CellChat, scTenifoldNet/scTenifoldKnk, virtual drug screening, and molecular docking were used for functional characterization. Results Fibroblasts were the dominant cellular source of the six hub genes in both ICM and LUAD. COL3A1 progressively increased along fibroblast pseudotime trajectories. Virtual overexpression and knockout analyses showed convergence on extracellular matrix organization, inflammatory response, and immune regulation. Cell communication analysis identified fibroblasts as major senders of COLLAGEN signaling, with recipient cells shifting from inflammatory monocytes under normal conditions to proliferating macrophages in ICM and tumor cells in LUAD. Molecular docking nominated Digoxin, Ergotamine, and Ledipasvir as candidate RUNX2-binding compounds with favorable predicted docking scores. Conclusions This study provides a single cell framework linking ICM and LUAD through fibroblast driven extracellular matrix remodeling and inflammation, and identifies potential molecular targets and therapeutic candidates for further validation.