Leveraging the mass balances of cellular metabolism to infer absolute concentrations from relative abundance metabolomics data

Abstract

AbstractMotivationAs the large-scale study of metabolites and a direct readout of a system’s metabolic state, metabolomics has significant appeal as a source of information for many metabolic modelling platforms and other metabolic analysis tools. However, metabolomics data are typically reported in terms of relative abundances, which precluding use with tools where absolute concentrations are necessary. While chemical standards can be used to determine the absolute concentrations of metabolites, they are often time-consuming to run, expensive, or unavailable for many metabolites. A computational framework that can infer absolute concentrations without the use of chemical standards would be highly beneficial to the metabolomics community.ResultsWe have developed and characterized MetaboPAC, a computational strategy that leverages the mass balances of a system to infer absolute concentrations in metabolomics datasets. MetaboPAC uses a kinetic equations approach and an optimization approach to predict the most likely response factors that describe the relationship between absolute concentrations and their relative abundances. We determined that MetaboPAC performed significantly better than the other approaches assessed on noiseless data when at least 60% of kinetic equations are known a priori. Under the most realistic conditions (low sampling frequency, high noise data), MetaboPAC significantly outperformed other methods in the majority of cases when 100% of the kinetic equations were known. For metabolomics datasets extracted from systems that are well-studied and have partially known kinetic structures, MetaboPAC can provide valuable insight about their absolute concentration profiles.