Network Pharmacology and Bioinformatic Prediction of Five Medicinal-Food Homologous Plants Against the Progression from Chronic Atrophic Gastritis to Gastric Cancer.

Chronic atrophic gastritis (CAG) progresses through distinct pathological stages to gastric cancer. This computational study employed network pharmacology and bioinformatics to predict the active compounds, targets, and pathways of five medicinal-food homologous plants (Houttuyniae Herba, Mume Fructus, Rhizoma Dioscoreae, Portulacae Herba, and Curcumae longae Rhizoma) against this progression. We identified 28 active compounds and 242 targets from TCMSP. differentially expressed genes from GSE130823 and GSE55696 were integrated with weighted gene co-expression network analysis, yielding 129 candidate targets. Machine learning (LASSO and random forest) prioritized AR, EIF6, GSTM2, and FOSB as potential diagnostic biomarkers distinguishing CAG from cancer. Enrichment analysis predicted involvement of IL-17, TNF, and p53 pathways. Compound-target-pathway network analysis identified quercetin and kaempferol as multi-target compounds. Molecular docking predicted binding affinities ranging from weak-to-moderate for core genes (-2.60 to -5.83 kcal/mol) and moderate-to-strong for inflammatory targets (up to -7.25 kcal/mol for quercetin-TNF). These findings computationally predict candidate mechanisms of these five plants, offering a rationale for future experimental validation and therapeutic development.