Integrative bioinformatics and deep learning to identify common genetic pathways in Crohn’s disease and ischemic cardiomyopathy

Abstract

Crohn’s disease (CD) and ischemic cardiomyopathy (ICM) share inflammatory characteristics, yet their common genetic underpinnings remain underexplored. Using an integrative bioinformatics approach, we analyzed GEO datasets (GSE3365 and GSE9128) to identify shared genetic pathways between CD and ICM. Through differential expression analysis, we identified 60 common differentially expressed genes (CDEGs). Functional enrichment analysis revealed enrichment in inflammatory pathways, including NF-κB and TNF-α signaling, highlighting their role in disease pathogenesis. We conducted microRNA (miRNA), transcription factor (TF), and protein–protein interaction (PPI) analyses to uncover regulatory networks. Notably, hsa-miR-98-5p emerged as a key miRNA, while RELA and NFKB1 were identified as prominent TFs interacting with CDEGs. Six hub genes—IL1B, CXCL8, CXCL2, TLR2, FCGR1A, and FCGR2A—were pinpointed, demonstrating high diagnostic potential via receiver operating characteristic (ROC) analysis. To advance diagnostic precision, we developed AutoClass, a deep learning framework that leverages hub gene regulatory networks to classify CD patients with approximately 95 % accuracy. These hub genes and regulators likely drive neutrophil-mediated inflammation, offering insights into the molecular interplay between CD and ICM. Our findings suggest that the identified CDEGs, miRNAs, and TFs hold promise as therapeutic targets and biomarkers, paving the way for precision medicine approaches in managing CD and its cardiovascular complications. Future experimental validation and cohort expansion could further elucidate these shared mechanisms, enhancing their translational impact.