A kinetic-aware approach to infer metabolic variations and flux using transcriptomics and metabolomics data

Estimating metabolic flux and its variations across various disease conditions poses significant challenges in metabolic research. These challenges stem from several fundamental limitations, including the requirement for disease and tissue-specific metabolic models, nonlinear enzyme kinetics linking enzyme and substrate alterations to reaction flux, and the unpaired, snap-shot nature of data collected across different omics modalities. Additionally, there is uncertainty in computational predictions.

To address these issues, researchers have developed Michaelis-Menten Model-Based Flux Estimation Analysis (mmFEA), a Monte Carlo framework designed to estimate condition-specific flux changes by integrating paired or unpaired metabolomics and transcriptomics (or proteomics) data. This innovative method breaks down each reaction-rate change into enzyme- and substrate-associated components, utilizing the Michaelis-Menten kinetics equation to assess reaction rates.

A key feature of mmFEA is the introduction of a baseline metabolite saturation rate, which is derived by integrating protein language model-predicted kinetic parameters and human baseline-level metabolic concentrations. This approach enables kinetic-aware integration of unpaired substrate and enzyme level measurements. The distribution of metabolic flux and variations between different conditions is computed using Markov Chain Monte Carlo (MCMC) sampling.

In this process, the Michaelis-Menten-derived marginal flux distribution serves as the prior, while coherency in flux balance acts as the likelihood.

To validate the efficacy of mmFEA, researchers generated an in-house multi-omics dataset comprising transcriptomics, metabolomics, metabolic activity functional assay, and CRISPR screening data from pancreatic cancer cell lines treated with APEX1 inhibitors. The results demonstrated that mmFEA could accurately capture experimentally observed metabolic changes and outperform all baseline methods.

This analysis highlighted the necessity of utilizing both substrate and enzyme modalities, along with a kinetic-aware model, for accurate metabolic flux assessment.

Further analysis using independent pancreatic cancer cohorts reinforced the robustness of mmFEA, underscoring the importance of integrating condition-linked unpaired data. The method has also shown promise in pan-cancer and spatial multi-omics applications, facilitating the resolution of context-dependent metabolic variations when direct flux measurements are not available.

Overall, mmFEA provides a mechanistically grounded framework for estimating relative metabolic flux changes and their associated uncertainties from heterogeneous omics data.