Metabolic flux analysis (MFA) plays a central role in metabolic engineering and systems biology [1]. Metabolic fluxes most closely reflect the underlying metabolic phenotype, whereas other 'omics approaches only yield a sense of metabolic capacities (transcriptomics/proteomics) or thermodynamic driving forces (metabolomics). Metabolic flux analysis is particular important in rational strain engineering, where we specifically seek to manipulate the metabolic phenotype.

Due to the high complexity of the examined metabolic network, flux analysis typically involves the use of a stoichiometric model, in which the metabolic reactions available to the cell are parameterized before the fluxes are estimated from experimental data [2]. State-of-art flux analysis today includes the use of stable isotopes to overcome problems such as incomplete resolution of important cellular pathways or the need to rely on stoichiometric parameters with high uncertainty such as ATP yield (Y_{x/ATP}) or P/O ratio which are inherently linked to the purely stoichiometric approaches [3]. ^{13}C-based MFA therefore is a powerful extension of MFA [3]. In such studies, after feeding ^{13}C-labelled substrate(s), one measures the ^{13}C tracer enrichment patterns of metabolites that are rich in flux information, using instruments such as nuclear magnetic resonance spectroscopy (NMR) [4, 5] or mass spectrometry (MS) [6]. There are mainly two different approaches to extract flux information from the labelling patterns: by model-based flux fitting [3], and by analytical interpretation of flux ratios [7] (both approaches briefly reviewed in [8]). Redundant pathways that contribute differently to tracer distribution can thus be resolved. Flux analysis is carried out independently from energy and redox balancing, because the balancing equations only involve the carbon backbone. Conversely, the flux results can be used to check the consistency of energy and redox balances [9].

There has been significant development especially concerning the experimental framework for ^{13}C MFA [10]. ^{13}C MFA has been applied to various prokaryotic and eukaryotic systems [11–13] involving miniaturized screening studies in small scale [14, 15]. There is an increasing trend towards large-scale network-based stationary ^{13}C MFA [16, 17], as well as non-stationary (i.e., dynamic) ^{13}C MFA [18–20]. Large-scale metabolic models are preferred in order to capture, as many reactions as possible, bearing effects on carbon labelling, and to maintain global consistency of flux estimates. Considering metabolism in isolated parts or using overly summarized metabolic models can lead to biased results [17]. However, specifying large sets of isotopomer balances and subsequently performing parameter estimation can be very cumbersome.

Several software packages have been developed to facilitate flux analysis, the most popular being FiatFlux [21] and 13C-FLUX [22]. FiatFlux implements the flux ratio approach to ^{13}C MFA [7] and comes preconfigured to derive flux ratios and net fluxes for [1-^{13}C]- and [U-^{13}C]-glucose experiments and GC-MS analysis of proteinogenic amino acids for several microorganisms. Recent developments allow to generate equation systems automatically [23], which facilitates the extension of the flux ratio approach to various metabolic models, input substrates and labelling data.

In contrast, 13C-FLUX is a general purpose package for modelling, simulation, design, evaluation, and statistical analysis of ^{13}C-labelling experiments [22]. Unfortunately, 13C-FLUX is relatively cumbersome to use in terms of requiring the user to specify free fluxes, to set up the initial solution, and to manually initialize and terminate each optimization. It is not possible to perform multiple rounds of optimization unsupervised, which is frequently used to check convergence of the optimization results. There is a general lack of support in aspects of experimental design, i.e., explore change in labelling patterns for a different flux distribution and/or various combinations of input substrates. For expert users, there is limited opportunity to modify source code for implementation of new algorithms and workflows for different labelling problems.

There is a need for ^{13}C MFA tool that is simple, flexible and transparent. Fast computation is crucial. For a non-expert user, the software must enable a smooth reproducible workflow covering the whole process from metabolic model definition to flux estimation. A flexible approach necessarily supports user-defined metabolic systems, while a transparent computational model offers expert users the opportunity to tailor make downstream algorithms for parameter estimation and statistical analysis.

To meet this challenge, we have developed OpenFLUX, a simple yet flexible application to perform steady-state ^{13}C MFA using mass isotopomer distribution data. OpenFLUX provides the user a versatile and intuitive spreadsheet-based interface to control the underlying metabolite and isotopomer balance models used for flux analysis and allowing for the implementation of large-scale metabolic networks. The user then has the option of using the accompanying algorithm package for flux estimation and sensitivity analysis, or applying alternative numerical approaches for flux analysis (e.g., [24]). OpenFLUX generates isotopomer balance model based on the EMU decomposition algorithm [25]. Using EMU variables is computationally more efficient because the number of necessary isotopomer balances is significantly reduced [25] compared to alternative representations of labelling distribution of metabolites, such as AAV (atom activity vector) [26], IDV (isotopomer distribution vector) [27], cumomer [28] and bondomer [29].

The present work describes the implementation and validation of OpenFLUX. Specifically, we explain the tasks performed by OpenFLUX, provide an illustration of the model definition setup of a hypothetical metabolic model, and also describe the structure and contents of the resulting metabolic models. The software is then validated by reproducing published ^{13}C MFA results.