To aid the integration of profiles from proteins, lipids and metabolites, we generated a network that combines a protein–protein interaction network and enzymatic and genetic interactions of proteins with metabolites and lipids. The hyperbolic embedding of the network was used to produce a map that reflects cellular functions and discriminates hubs. Proximity in the map can be used to find relations across proteins, lipids and metabolites through a user-friendly Shiny R software package (Omint) [1]. We demonstrate the use of the multiomic network to identify lipids and metabolites associated with CVD-related proteins and to study the temporal effects of the antidiabetic drug empagliflozin on lipid metabolism using data from the EmDia study [2]. The EmDia study demonstrated that some of the positive effects of the drug are related to lipid modulation.
To correctly identify lipidomics and metabolomics data points in liquid chromatography, one needs to know the retention times (RTs) of the corresponding compounds, but experimental data is scarce and limited to the compounds already tested. It would be therefore useful to predict RTs for metabolites based on their molecular structure. We show that it is possible to do this by training a model on peptide RT data, which is abundant [3]. The method uses the ChemBERTa language model trained on peptides and performs better if it is made to predict RDKit molecular descriptors (physical, chemical and structural) in addition to RTs. Predictions improve significantly if some known lipid RT data is added to the peptide training set (even just using 5% of the available lipid RT data), suggesting that the method can be incrementally improved with heterogeneous experimental RT data.
References
[1] Anyaegbunam, U.A., A. Vagiona, V. ten Cate, K. Bauer, T. Schmidlin, U. Distler, S. Tenzer, E. Araldi, L. Bindila, P. Wild and M.A. Andrade-Navarro. 2025. A map of the lipid-metabolite-protein network to aid multi-omics integration. Biomolecules. 15, 484.
[2] Bauer, K., D. Baker, R. Lerner, T. Koeck, G. Buch, Z. Fischer, E.E. Esenkova, M. Nuber, M.A. Andrade-Navarro, S. Tenzer, P.S. Wild, L. Bindila, E. Araldi. 2025. Effect of Empagliflozin on the plasma lipidome in patients with type 2 diabetes mellitus – results from the EmDia clinical trial. Cardiovasc. Diabetol. 24, 359.
[3] Anyaegbunam, U.A., D. Teschner, T. Schmidlin, A. Hildebrandt, J.U. Mayer, M. Sprang, M.A. Andrade-Navarro. 2026. Cross-domain transfer learning from peptides to metabolites using a multi-property fine-tuned LLM. Bioinformatics. 42, btag493.