Lemonite: identification of regulatory metabolites through data-driven, interpretable integration of transcriptomics and metabolomics data

Lemonite identifies regulatory metabolites linked to coexpressed gene modules in glioblastoma and inflammatory bowel disease, including myo-inositol and phosphatidylcholines, with over 50 coherent modules per disease.

• Why it matters: Metabolites are often missing from genome-wide gene regulatory network inference, and existing integration methods are limited by weak interpretability or reliance on incomplete prior knowledge.
• What they did: Lemonite jointly infers module networks from bulk transcriptomics and metabolomics without requiring prior differential analysis or complete metabolite annotation, using a knowledge graph with over 370,000 metabolite-gene links and 2.1 million protein-protein interactions.
• The result: In glioblastoma, myo-inositol and phosphatidylcholines with IRF6 are linked to mesenchymal-like immune programs enriched in tumor-associated macrophages and monocytes, while in inflammatory bowel disease, prioritized metabolites are linked to in vitro expression changes of predicted targets.

AI summary of the abstract · may contain errors · not medical advice · read the paper.