OpenFLUX2: 13 C-MFA modeling software package adjusted for the comprehensive analysis of single and parallel labeling experiments

Abstract

BackgroundSteady-state 13C-based metabolic flux analysis (13C-MFA) is the most powerful method available for the quantification of intracellular fluxes. These analyses include concertedly linked experimental and computational stages: (i) assuming the metabolic model and optimizing the experimental design; (ii) feeding the investigated organism using a chosen 13C-labeled substrate (tracer); (iii) measuring the extracellular effluxes and detecting the 13C-patterns of intracellular metabolites; and (iv) computing flux parameters that minimize the differences between observed and simulated measurements, followed by evaluating flux statistics. In its early stages, 13C-MFA was performed on the basis of data obtained in a single labeling experiment (SLE) followed by exploiting the developed high-performance computational software. Recently, the advantages of parallel labeling experiments (PLEs), where several LEs are conducted under the conditions differing only by the tracer(s) choice, were demonstrated, particularly with regard to improving flux precision due to the synergy of complementary information. The availability of an open-source software adjusted for PLE-based 13C-MFA is an important factor for PLE implementation.ResultsThe open-source software OpenFLUX, initially developed for the analysis of SLEs, was extended for the computation of PLE data. Using the OpenFLUX2, in silico simulation confirmed that flux precision is improved when 13C-MFA is implemented by fitting PLE data to the common model compared with SLE-based analysis. Efficient flux resolution could be achieved in the PLE-mediated analysis when the choice of tracer was based on an experimental design computed to minimize the flux variances from different parts of the metabolic network. The analysis provided by OpenFLUX2 mainly includes (i) the optimization of the experimental design, (ii) the computation of the flux parameters from LEs data, (iii) goodness-of-fit testing of the model’s adequacy, (iv) drawing conclusions concerning the identifiability of fluxes and construction of a contribution matrix reflecting the relative contribution of the measurement variances to the flux variances, and (v) precise determination of flux confidence intervals using a fine-tunable and convergence-controlled Monte Carlo-based method.ConclusionsThe developed open-source OpenFLUX2 provides a friendly software environment that facilitates beginners and existing OpenFLUX users to implement LEs for steady-state 13C-MFA including experimental design, quantitative evaluation of flux parameters and statistics.Electronic supplementary materialThe online version of this article (doi:10.1186/s12934-014-0152-x) contains supplementary material, which is available to authorized users.

The JBEI quantitative metabolic modeling library (jQMM): a python library for modeling microbial metabolism

“…The jQMM library is also the first one to provide a method to calculate fluxes for genome-scale models constrained by 13 C labeling data through 2S- 13 C MFA, and is the first to unify the FBA, 13 C MFA and 2S- 13 C MFA in a single package. Unlike METRAN [27], INCA [28] and 13CFLUX2 [29]), it is open source and, unlike OPENFLUX [30] and OPENFLUX2 [31], it is written in Python. Furthermore, it includes an example on how to calculate fluxes for microbial communities through 13 C MFA, and the tools to leverage other types of -omics data (e.g.…”

“…For 13 C MFA there exists some closed-source packages (e.g. METRAN [27], INCA [28], 13CFLUX2 [29]), a few open-source packages (OPENFLUX [30] and OPENFLUX2 [31]) but no open source python-based libraries. For 2S- 13 C MFA, jQMM is the first public library.…”

Mapping Salmonella typhimurium pathways using 13C metabolic flux analysis

“…Successive reactions, without changes in atom transition, were lumped. The model was written according to the OpenFLUX instructions and examples (Quek et al, 2009;Quek and Nielsen, 2014;Shupletsov et al, 2014).…”

“…Metabolic flux values were estimated using the MATLAB-based modeling software OpenFLUX2 (Shupletsov et al, 2014). Fluxes were computed 10 times, with 500 iterations, starting with random initial values.…”

To be certain about the uncertainty: Bayesian statistics for13C metabolic flux analysis

“…PL-CoIs, on the other hand, are dependent on that each optimization that is performed when calculating the CoI bounds successfully discovers the global optimum. It is observed by Shupletsov et al (2014) that a single optimization run that terminates at a non-optimal point influences the PL-CoIs directly. Heuristic measures for making PL more robust against errors arising from optimization can be undertaken, for example, in the form of multi-start procedures or by applying exhaustive grid-search (the latter being computationally feasible only for low dimensional problems).…”