Inverse problem in human metabolism: insights from hybrid data-driven and mechanistic approaches

This thesis advances precision medicine through a systems-level investigation of human metabolism, integrating multi-omics profiling, organ-level modeling, and computational simulation. By connecting molecular measurements to mechanistic understanding, it aims to predict disease progression and therapeutic responses. The first project examined glycogen storage disease type 1a (GSD1a), caused by G6PC1 mutations. Multi-omics profiling of patient serum identified distinct proteomic and metabolomic signatures reflecting disruptions in gluconeogenesis, TCA cycle activity, and choline metabolism. Liver-enriched proteins, including ALDOB, correlated with clinical injury markers, revealing potential biomarkers for hepatic stress. Network analyses highlighted systemic metabolic dysregulation, emphasizing correlative associations between molecular and clinical phenotypes. The second project developed multi-regional gastric assembloids (MRAs) to model human stomach physiology in vitro. MRAs integrated region-specific organoids into tubular constructs, preserving fundic, body, and antral specialization. Transcriptomic and ligand–receptor analyses confirmed inter-regional communication through key signaling pathways. Functionally, MRAs exhibited histamine-induced acid secretion inhibited by omeprazole, and protein analyses confirmed parietal cell differentiation, establishing a platform for studying gastric disease, drug response, and host–pathogen interactions. The third project introduced a computational pipeline integrating proteomic and metabolomic data into genome-scale metabolic models. Applied to methylmalonic acidemia, it predicted context-specific metabolic fluxes and rewiring, particularly in the TCA cycle, mitochondrial transport, and amino acid handling, providing mechanistic insights into compensatory adaptations. Additional work explored liver extracellular matrix remodeling across fibrosis, development, and cancer, revealing conserved programs and a desmoplastic ECM signature in aggressive colorectal cancer, along with CRISPR-based screens identifying genes controlling directional fibroblast migration. Together, these studies unify molecular profiling, tissue engineering, and computational modeling into a systems-medicine framework, transforming descriptive biology into predictive science and advancing precision medicine toward mechanism-based, clinically actionable insights.