Network-guided computational prioritization of gut microbiota-associated metabolites in sepsis-associated intestinal injury.

Abstract

Sepsis-associated intestinal injury involves barrier dysfunction and altered microbial metabolism, but microbiota-associated metabolites have not been systematically prioritized. We integrated curated microbe-metabolite and host-gene records, ligand-based target prediction, disease-gene databases, protein-protein interaction analysis, enrichment, exploratory transcriptomics, ADME/toxicity predictions, and a uniform three-component ranking. The intersection of 1518 predicted targets, 154 gutMGene-supported host targets, and 4402 disease-associated genes yielded 63 genes. GSE202261 included two CLP and two sham intestinal samples, and no hub-gene or pathway result remained significant after FDR correction; it was therefore treated as exploratory. The archived AutoDock4 files lacked sufficient preparation and validation records and were excluded from ranking. In a separate structural assessment, both AKT1 structures failed native-ligand redocking. IL1B structure 8C3U passed redocking, but a literature-derived activity-labelled control panel failed one of five directional tests (AUROC 0.80); candidate docking was therefore not undertaken. Among 19 identity-deduplicated metabolites, 3-(4-HPP) ranked first (F₃=0.814), I3LA second (0.600), and 3-IPA third (0.580). 3-(4-HPP) remained first across 5000 admissible weight schemes. These results prioritize candidates for experimental testing but do not establish binding or biological efficacy.