Background
Metabolic dysfunction–associated steatotic liver disease (MASLD) affects a rapidly growing population, yet clinical decision-making still relies on limited biomarkers and 2D biopsy reads. Recent single-cell and spatial omics map how hepatocytes and non-parenchymal cells vary from portal to central vein (“zonation”), but most analyses remain descriptive and siloed. This project will turn those measurements into a Digital Liver Twin—a mechanistic, personalised simulation of hepatic metabolism—by combining spatial multi-omics with genome-scale metabolic models (GSMMs) and relevant clinical data.
Novelty & Importance
The Digital Liver Twin will be the first spatially informed, personalised GSMM of human liver that (i) encodes portal–central zonation as quantitative constraints, and (ii) couples extrahepatic inputs from adipose and muscle to reflect whole-body metabolism. By unifying mechanistic simulation with machine-learning readouts, the twin aims to predict individual disease progression and drug or metabolic stimulus responses. This approach directly supports EPSRC’s “Complex Simulations & Digital Twins” theme by providing a transparent, testable virtual patient for MASLD—accelerating hypothesis generation, improving patient stratification, and guiding therapeutic decisions while reducing laboratory burden.
Aims & Objectives
1. Integrate multi-omics and clinical data: harmonise spatial/bulk/single-cell liver datasets with matched adipose/muscle context and clinical covariates.
2. Build the twin: construct a zonation-aware, personalised GSMM and implement extrahepatic exchange constraints; run baseline and stimulated (insulin, glucagon, fatty acids) simulations.
3. Validate predictions: compare simulated fluxes with tracer-based proxies; test pathway-level perturbations in organoids or precision-cut liver slices.
4. Predict progression and response: train models that combine twin outputs with longitudinal clinical data to yield an interpretable risk score and phenotype-specific responder signatures.
5. Translate to practice: package the twin as a reproducible software tool producing clinician-readable reports for decision support and study design.