HUman MEtabolic SimulationS

The HUMESS (HUman MEtabolic SimulationS) project aims to reconstruct a human metabolic network (characteristic of an individual metabolic phenotype) and automatically identify essential metabolic reactions that explain the phenotype, based on transcriptomic data.

The project is based on the expertise of LS2N in relation to the integration of heterogeneous data and multi-scale modeling. The ComBi]( team proposes to formalize the reconstruction of the metabolic network as a constraint-based problem (MILP). This problem can be solved using dedicated AI solvers (i.e., CPLEX). Solving the MILP problem will identify a metabolic network covering all genes identified by transcriptomics while being consistent with thermodynamic and mass action laws.

The network, which summarizes phenotypic activity based on gene expression, as well as dependencies between metabolic reactions, can then be explored using another formulation of a constraint-based problem (multi-objective linear problem — MOLP) to identify essential reactions for maintaining the phenotype (via a “bensolve” solver).

This project is funded for 2 years by Biogenouest.