A Network Pharmacology-Based Computational Workflow for Investigating Gut Microbiota Metabolites Associated with Coronary Atherosclerosis

The gut microbiota and its metabolites are closely associated with the pathogenesis of coronary atherosclerosis (CAS). However, a standardized computational pipeline for systematically integrating microbial, metabolite, and host target information remains lacking. Here, we established an integrative workflow combining multi-database mining, network pharmacology, and molecular docking to explore potential molecular links between gut microbial metabolites and CAS. Gut microbial metabolites and host target genes were retrieved from gutMGene, with additional metabolite targets predicted using SEA and SwissTargetPrediction. CAS-related targets were compiled from GEO, GeneCards, CTD, OMIM, and Open Targets. Multi-level intersection analysis yielded candidate targets, followed by construction of a GUT-TARGET-CAS network. Protein-protein interaction networks were analyzed via STRING and Cytoscape, and a Microbiota-Substrate-Metabolite-Target (M-S-M-T) network was built. Functional and disease enrichment analyses (GO, KEGG, and Disease Ontology) were performed, and drug-likeness and toxicity of candidate metabolites were assessed using SwissADME and ADMETlab 3.0. Finally, CB-Dock2 was employed to evaluate predicted structural compatibility between candidate metabolites and highly connected targets. Representative analysis identified 226 gut microbial metabolites, 1,518 predicted metabolite targets, and 2,462 CAS-related targets, from which 52 candidate targets were prioritized. The PPI network prioritized TNF, IL6, AKT1, TP53, IL1B, and PPARG as highly connected targets. The M-S-M-T network linked these targets with 53 metabolites, 33 substrates, and 82 gut microbial taxa; all edges represent database-derived associations rather than proven causal regulatory relationships. Lipinski-type physicochemical criteria and ADMET (absorption, distribution, metabolism, excretion, and toxicity) predictions were used as a developability-oriented prioritization step, whereas docking scores were used only for internal candidate ranking. This computational framework enables systematic integration of gut microbiota-metabolite-host target information and prioritizes candidate metabolites and molecular targets for subsequent experimental validation.