iMS2Flux has been designed to act as a high-throughput framework for MS data analysis, targeting MFA as its primary application, but is not inherently limited to MFA. The software is designed to be modular and flexible emphasizing a standard data exchange format. The standard data format allows for easy access through any spreadsheet application, and is supported with import and export modules to easily allow new tools to make use of the data. In Figure 2 an MFA workflow maximizing automation is illustrated, utilizing iMS2Flux to branch from data extraction to data analysis with 13C-Flux software (see ‘Supporting Information’). iMS2Flux defines not only the generic correction tool, but a fully standardized data format, and through it an automated workflow connecting third party extraction and analysis tools.
The performance of iMS2Flux was tested on a commercially available PC. The Perl interpreter was ActivePerl v.5.12.2 (from ActiveState). iMS2Flux is a non-threaded application and ran entirely on a single core. A set of GC-MS generated data comprising 128 chromatograms with a total of 65 fragments corresponding to 412 masses was processed in 119 seconds. To perform the benchmark iMS2Flux was set to check for missing data, detector threshold limits and poor peak values, extract an additional measurement (M+numC+1) from each fragment, perform natural abundance correction, generate the carbon labeling summary (post correction data check), generate average and standard deviations over replicates, and generate a complete set of output data (raw measurement, corrected measurement, average and standard deviation of corrected, and model data for each experimental set in 13CFlux FTBL format for inclusion in the MASS_SPECTROMETRY section). The data was pre-screened to ensure that the MS data would pass all data checks to complete processing.
As illustrated in Table 1
iMS2Flux offers a variety of options for data correction. Similar to MSCorr, it offers checks to ensure the MS data is within the upper and lower boundaries of the MS detector, whereas the tool CORRECTOR assumes the process MS data is accurate. Depending on the tools used to extract the relevant MS data from chromatograms, e.g. [30, 31, 42] or manufacturer software, checks for data accuracy and quality can be performed during data extraction. OpenFLUX is an MFA analysis tool that also provides a NOI correction tool (not directly integrated). Similar to MSCorr, the OpenFLUX correction is provided as a function in MATLAB (corrMatGen) which requires the user to enter the chemical formula and other specifics about each compound individually. MSCorr, corrMatGen, and CORRECTOR correct for NOIs, iMS2Flux allows additional correction methods: for original biomass as well as proton-loss or gain. Furthermore, iMS2Flux is capable of performing all the above mentioned corrections on large and heterogeneous data sets, comprising multiple analytes with multiple MDVs in multiple chromatograms. The addition of new analyte sum formulas in iMS2Flux is intuitive, since it only requires the total chemical formula of the new analyte, without separation of the metabolite derived part of an analyte from any derivatization reagent additions. Alternatively fully generalized analyte classes supporting multi-stage and multiple alternative derivatization are also possible. Finally the output of iMS2Flux is ready-to-use in MFA-dedicated software. The aforementioned FiatFlux is able to correct GC-MS data for natural abundance and original unlabeled biomass. The quality of the extracted MS data is checked in a similar way as in MSCorr, and faulty MS data can be removed manually from further calculations. Similar to MSCorr new compounds require a separation of the atoms of the analyte from the derivatizing agent. FiatFlux is focused on deriving flux ratios and absolute fluxes for microorganisms solely from 1-13C and/or U-13C glucose experiments combined with GC-MS analysis of amino acids .
Although iMS2Flux was designed to serve the needs of MFA, it can be used as a general tool to quantify stable isotope labeling in any kind of isotope tracer experiment, e.g. [32, 40, 43]. Furthermore, although carbon labeling with 13C is the method of choice in MFA, other elements such as nitrogen, hydrogen or oxygen are conceivable for tracer studies [44–46]. iMS2Flux can easily be adapted to any other element as isotope tracer. In order to allow the general application of iMS2Flux in MFA, independent of the MS platform the labeling data were acquired on, it was designed to process GC-MS, LC-MS or MS/MS data. Additionally, besides data from steady state labeling experiments, iMS2Flux can process dynamic labeling data as well. For the exploitation of the full potential of dynamic labeling experiments, such as short labeling time [47, 48], it is necessary to be able to measure and evaluate MS data not only derived from metabolic end products (storage compounds) but from metabolic intermediates, which can have a very fast turnover [49, 50]. This would increase the resolution of a metabolic network and can resolve precursor-product relationships which are difficult or impossible to resolve with data derived from end product labeling . To give iMS2Flux this capability, data of the elemental composition of polar soluble intermediates of primary metabolism, as previously published [32, 41], were included. This list of supported analytes can be extended as needed, in case new metabolites are of interest or a different derivatization strategy is applied.
In the context of measuring complex biological matrices of soluble metabolic intermediates, similar to metabolic profiling measurements, it seems appropriate to use specialized software. Since there are multiple software solutions available, especially dedicated to the alignment of multiple MS chromatograms and extracting the relevant MS data, e.g. [30, 31, 42] or manufacturer software, our efforts focused on finding a general input format that supports the respective data outputs, the TSV format described above. iMS2Flux was implemented in PERL which is freely available and runs on all major computing platforms. Furthermore, MS data are usually provided in tabular form, which is either already in the TSV format, or is easily exported to TSV, thus a text manipulation language was the obvious choice. PERL supports multiple programming paradigms and no compilers are required, as it is a dynamic language a respective script just needs to be edited and can be run directly. To further promote the use of iMS2Flux, the code is provided in full and since the program is not compiled the source is immediately available to be reviewed and extended for individual needs. To support flexibility the different data formats, optional data checks, data correction and output formats are contained in individual modules.