Integrated Network Pharmacology and Machine Learning Identify Potential Molecular Targets of Bayogenin Associated with Energy-Negative-Balance-Related Fatty Liver in Dairy Cows

Fatty liver in transition dairy cows is associated with negative energy balance and impaired hepatic adaptation. Bayogenin, a triterpenoid sapogenin associated with alfalfa saponins, has an unclear role in bovine hepatic metabolism. This study integrated published transcriptomic and metabolomic data from bovine hepatocytes treated with trimethylamine N-oxide (TMAO), a gut microbiota-associated metabolite, with machine-learning feature selection, network pharmacology, molecular docking, and molecular dynamics (MD) simulations. The input comprised 4383 differentially expressed genes and 66 metabolite features. The results showed that: (1) Boruta, Elastic net, and SVM-RFE prioritized nine putatively annotated metabolite features; four metabolic pathways had nominal p values below 0.05, but none remained significant after multiple-testing correction; (2) the integration of predicted Bayogenin targets, fatty liver-associated genes, human orthologs of DEGs, and predicted metabolite-associated targets yielded 33 candidate genes, with JAK2, MMP9, EGFR, and NR3C1 identified as provisional candidates based on network analysis; (3) the JAK2-Bayogenin and MMP9-Bayogenin complexes yielded the most favorable predicted docking scores and were each examined in a single 100-ns MD simulation. In conclusion, these computational findings generate testable hypotheses for future studies but do not establish Bayogenin-induced metabolic effects or direct target engagement.