Metabolic flux configuration determination using information entropy.

Area of Science:

• Systems Biology
• Metabolic Engineering
• Computational Biology

Background:

• Constraint-based models define metabolic flux solution spaces using mass balances.
• Redundant pathways and metabolic loops lead to multiple alternative solutions, causing ambiguity.
• Existing methods like flux sampling and economy of fluxes assumption (EFA) have limitations, including susceptibility to artifacts and questionable universal validity.

Purpose of the Study:

• To develop a novel constraint-based approach, MaxEnt, based on the principle of maximum entropy.
• To improve the accuracy and reliability of metabolic flux predictions.
• To address the ambiguity in metabolic flux configurations and reduce reliance on potentially invalid assumptions.

Main Methods:

• Formulated the MaxEnt approach using the principle of maximum entropy.
• Applied MaxEnt to publicly available flux data from Escherichia coli and Saccharomyces cerevisiae.
• Compared MaxEnt predictions against flux sampling and EFA-based methods.
• Evaluated method performance under varying levels of overflow metabolism.

Main Results:

• MaxEnt achieved mean square errors (MSE) three orders of magnitude lower than flux sampling.
• MaxEnt and flux sampling correctly predicted flux through E. coli's glyoxylate cycle, unlike EFA-based methods.
• MaxEnt's accuracy remained unaffected by overflow metabolism, while EFA-based methods showed decreased performance.
• MaxEnt demonstrated reduced sensitivity to thermodynamically infeasible cycles and less susceptibility to overfitting compared to EFA.

Conclusions:

• MaxEnt provides a more accurate and robust method for metabolic flux prediction.
• The maximum entropy principle offers a reliable framework for resolving metabolic ambiguities.
• MaxEnt outperforms existing methods in accuracy, handling of specific metabolic pathways, and robustness to overflow metabolism.