Abstract
Isotope labeling networks (ILNs) are graphs explaining the flow of isotope labeled molecules in a metabolic network. Moreover, they are the structural backbone of metabolic flux analysis (MFA) by isotopic tracers which has been established as a standard experimental tool in fluxomics. To configure an isotope labeling experiment (ILE) for MFA, the structure of the corresponding ILN must be understood to a certain extent even by a practitioner. Graph algorithms help to analyze the network structure but produce rather abstract results. Here, the major obstruction is the high dimension of these networks and the large number of network components which, consequently, are hard to figure out manually. At the interface between theory and experiment, the three-dimensional interactive visualization tool CumoVis has been developed for exploring the network structure in a step by step manner. Navigation and orientation within ILNs are supported by exploiting the natural 3D structure of an underlying metabolite network with stacked labeled particles on top of each metabolite node. Network exploration is facilitated by rotating, zooming, forward and backward path tracing and, most important, network component reduction. All features of CumoVis are explained with an educational example and a realistic network describing carbon flow in the citric acid cycle.
Similar content being viewed by others
Abbreviations
- ILN:
-
Isotope labeling network
- MFA:
-
Metabolic flux analysis
- ILE:
-
Isotope labeling experiment
- CumoVis:
-
Cumomer network visualization tool
- M:
-
Metabolite symbols
- M#101:
-
Isotopomer notation (M, metabolite symbol; 0, unlabeled; 1, labeled)
- M#1X1:
-
Cumomer notation (M, metabolite symbol; X, don’t care; 1, labeled)
- M#abc :
-
Enumeration of atom positions
- v: A→B:
-
Reaction notation
- 3D:
-
Three dimensional
- WL:
-
Weight level
- CC:
-
Connected component
- SCC:
-
Strongly connected component
- DAG:
-
Directed acyclic graph
- AcCoA:
-
Acetyl-CoA
- AcN:
-
cis-Aconitate
- AKG:
-
Alpha-Ketoglutarate
- Cit:
-
Citrate
- CO2:
-
Carbon dioxide
- Fum:
-
Fumarate
- GlyOx:
-
Glyoxylate
- ICit:
-
Isocitrate
- Mal:
-
Malate
- OAA:
-
Oxaloacetate
- Pyr:
-
Pyruvate
- Succ:
-
Succinate
- SucCoA:
-
Succinyl-CoA
References
Wiechert W (2001) 13C metabolic flux analysis. Metab Eng 3:195–206
Sauer U (2006) Metabolic networks in motion: 13C-based flux analysis. Mol Syst Biol 2:62
Szyperski T (1998) 13C -NMR, MS and metabolic flux balancing in biotechnology research. Q Rev Biophys 31:41–106
Sauer U (2004) High-throughput phenomics: experimental methods for mapping fluxomes. Curr Opin Biotechnol 15:58–63
Marx A, Graaf AAd, Wiechert W, Eggeling L, Sahm H (1996) Determination of the fluxes in central metabolism of Corynebacterium glutamicum by NMR spectroscopy combined with metabolite balancing. Biotechnol Bioeng 49:111–129
Ratcliffe RG, Shachar-Hill Y (2006) Measuring multiple fluxes through plant metabolic networks. Plant J 45:490–511
Winden WAv, Gulik WMv, Schipper D, Verheijen PJT, Krabben P, Vinke JL, Heijnen JJ (2003) Metabolic flux and metabolic network analysis of Penicillium chrysogenum using 2D [13C, 1H] COSY NMR measurements and cumulative Bondomer simulation. Biotechnol Bioeng 83:75–92
Kelleher JK (2004) Probing metabolic pathways with isotopic tracers: insights from mammalian metabolic physiology. Metab Eng 6:1–5
Malloy CR, Sherry AD, Jeffrey FMH (1990) Analysis of tricarbocylix acid cycle of the heart using 13C isotope isomers. Am J Physiol 259:987–995
Drysch A, El Massaoudi M, Wiechert W, de Graaf AA, Takors R (2004) Serial flux mapping of Corynebacterium glutamicum during fed-batch L-lysine production using the sensor reactor approach. Biotechnol Bioeng 85:497–505
Fischer E, Sauer U (2005) Large-scale in vivo fluxes reveal rigidity and suboptimal performance of B. subtilis metabolism. Nat Genet 37:636–640
Blank LM, Kuepfer L, Sauer U (2005) Large-scale 13C-flux analysis reveals mechanistic principles of metabolic network robustness to null mutations in yeast. Genome Biol 6:R49
Strohhäcker J, Graaf AAd, Schoberth SM, Wittig RM, Sahm H (1993) 31P nuclear magnetic resonance studies of ethanol inhibition in Zymomonas mobilis. Arch Microbiol 159:484–490
Tesch M, Graaf AAd, Sahm H (1999) In vivo fluxes in the ammonium-assimilatory pathways in Corynebacterium glutamicum studied by 15N nuclear magnetic resonance. Appl Environ Microbiol 65:1099–1109
Antoniewicz MR, Kelleher JK, Stephanopoulos G (2007) Elementary metabolite units (EMU): a novel framework for modeling isotopic distributions. Metab Eng 1:68–86
Christensen B, Nielsen J (1999) Isotopomer analysis using GC-MS. Metab Eng 1:282–290
Graaf AAd (2000) Use of 13C labelling and NMR spectroscopy in metabolic flux analysis, Chap. 4. In: Barbotin J-N, Portais J-C (eds) NMR in biotechnology: theory and applications, Horizon Scientific Press, Wymondham
Wiechert W (2002) An introduction to 13C metabolic flux analysis. Genet Eng Princ Methods 24:215–238
Wiechert W, Möllney M, Isermann N, Wurzel M, de Graaf AA (1999) Bidirectional reaction steps in metabolic networks. Part III: Explicit solution and analysis of isotopomer labeling systems. Biotechnol Bioeng 66:69–85
Möllney M, Wiechert W, Kownatzki D, de Graaf AA (1999) Bidirectional reaction steps in metabolic networks. Part IV: Optimal design of isotopomer labeling experiments. Biotechnol Bioeng 66:86–103
Schmidt K, Carlsen M, Nielsen J, Villadsen J (1997) Modelling isotopomer distribution in biochemical networks using isotopomer mapping matrices. Biotechnol Bioeng 55:831–840
Fischer E, Zamboni N, Sauer U (2004) High-throughput metabolic flux analysis based on gas chromatography: mass spectrometry derived 13C constraints. Anal Biochem 325:308–316
Rantanen A, Mielikäinen T, Rousu J, Maaheimo H, Ukkonen E (2006) Planning optimal measurements of isotopomer distributions for estimation of metabolic fluxes. Bioinformatics 22:1198–1206
Winden Wv, Heijnen JJ, Verheijen PJT (2002) Cumulative bondomers: a new concept in flux analysis from 2D [13C,1H] COSY data. Biotechnol Bioeng 80:731–745
Klapa MI, Park SM, Sinskey AJ, Stephanopoulos G (1999) Metabolite and isotopomer balancing in the analysis of metabolic cycles: I. Theory Biotechnol Bioeng 62:375–391
Weitzel M, Nöh K, Wiechert W (2007) The topology of metabolic carbon labeling networks. BMC Bioinf 8:315
Katz J, Wals P, Lee W-NP (1993) Isotopomer studies of gluconeogenesis and the Krebs cycle with 13C-labeled lactate. J Biol Chem 268:25509–25521
Wiechert W, Möllney M, Petersen S, de Graaf AA (2001) A universal framework for 13C metabolic flux analysis. Metab Eng 3:265–283
Dauner M, Sauer U (2000) GC-MS analysis of amino acids rapidly provides rich information for isotopomer balancing. Biotechnol Prog 16:642–649
Nöh K, Wahl A, Wiechert W (2006) Computational tools for isotopically instationary 13C labelling experiments under metabolic steady state conditions. Metab Eng 8:554–577
Wiechert W, Siefke C, Graaf AAd, Marx A (1997) Bidirectional reaction steps in metabolic networks. Part II: Flux estimation and statistical analysis. Biotechnol Bioeng 55:118–135
Zamboni N, Fischer E, Sauer U (2005) FiatFlux: a software for metabolic flux analysis from 13C-glucose experiments. BMC Bioinf 6:209
Klapa MI, Park SM, Sinskey AJ, Stephanopoulos G (1999) Metabolite and isotopomer balancing in the analysis of metabolic cycles: I. Theory Biotechnol Bioeng 62:375–391
Park SM, Shaw-Reid C, Sinskey AJ, Stephanopoulos G (1997) Elucidation of anaplerotic pathways in Corynebacterium glutamicum via 13C-NMR spectroscopy and GC-MS. Appl Microbiol Biotechnol 47:430–440
Jeffrey FMH, Rajagopal A, Malloy CR, Sherry AD (1991) 13C-NMR: a simple yet comprehensive method for analysis of intermediary metabolism. TIBS 16:5–10
Sriram G, Shanks JV (2004) Improvements in metabolic flux analysis using carbon bond labeling experiments: bondomer balancing and Boolean function mapping. Metab Eng 6:116–132
Forbes NS, Clark DS, HW Blanch (2001) Using isotopomer path tracing to quantify metabolic fluxes in pathway models containing reversible reactions. Biotechnol Bioeng 74:196–211
Sherry AD, Jeffrey FMH, Malloy CR (2004) Analytical solutions for 13C isotopomer metabolic conditions: substrate oxidation, multiple andgluconeogenesis. Metab Eng 6:12–24
Donato LD, Rosiers CD, Montgomery JA, David F, Garneau M, Brunengraber H (1993) Rates of gluconeogenesis and citric acid cycle in perfused livers, assessed from the mass spectrometric assay of the 13C labeling pattern of glutamate. J Biol Chem 268:4170–4180
Arita M (2005) Introduction to the ARM database: database on chemical transformations in metabolism for tracing pathways, In: Nishioka MtaT (eds) Metabolomics: the frontier of systems biology, Springer Tokyo, Tokyo
Arita M, Fujiwara Y, Nakanishi Y (2006) Map editor for the atomic reconstruction of metabolism (ARM), in plant metabolomics. Springer, Berlin
Brandes U, Dwyer T, Schreiber F (eds) (2003) Visualizing related metabolic pathways in two and a half dimensions. Springer lecture notes in computer science, vol 11. International symposium on graph drawing, Springer, Berlin
Wiechert W, Wurzel M (2001) Metabolic isotopomer labeling systems. Part I: Global dynamic behaviour. Math Biosci 169:173–205
Jünger M, Mutzel P (2003) Graph drawing software. Springer, New York
Karp P, Paley SM (1994) Automated drawing of metabolic pathways. In: Third international conference on bioinformatics and genome research
Rost U, Kummer U (2004) Visualisation of biochemical network simulations with SimWiz. IEE Syst Biol 1:184–189
Noack S, Wahl A, Qeli E, Freisleben B, Wiechert W (2007) Visualizing regulatory interactions in metabolic networks. BMC Biol 5:46
Wegner K (2005) SimWiz3D: visualising biochemical simulation results. In: Medical information visualisation: biomedical visualisation, 2005 (MediVis 2005). Proceedings. Third International Conference on 77–82
Koike H (1993) The role of another spatial dimension in software visualization. ACM Trans Inf Syst 11:266–286
Cohen RF, Eades P, Lin T, Ruskey F (1995) Three-dimensional graph drawing. In:Tamassia R, Tollis IG (eds) Proceedings graph drawing, pp 1–11
Frati F, Battista GD (2007) Three dimensional drawings of bounded degree trees. In: Kaufmann M, Wagner D (eds) Proceedings graph drawing, pp 89–94
Ho J, Hong S-H (2006) Drawing clustered graphs in three dimensions. In: Healy P, Nikolov NS (eds) Proceedings graph drawing, pp 492–502
Dickerson JA, Yang Y, Blom K, Reinot A, Lie J, Cruz-Neira C, Wurtele ES (2004) Using virtual reality to understand complex metabolic networks. In: Atlantic symposium comp biol genomic info systems technol
Hong S-H (2006) MultiPlane: a new framework for drawing graphs in three dimensions. In: Graph drawing, pp 414–415
Brandes U, Corman S (2002) Visual unrolling of network evolution and the analysis of dynamic discourse. In: IEEE symposium on information visualization (INFOVIS ‘02), pp 145–151
Wiechert W (2001) 13C metabolic flux analysis. Metab Eng 3:195–206
Wiechert W, Graaf AAd (1996) In vivo stationary flux analysis by 13-C labelling experiments. Adv Biochem Eng Biotechnol 54:109–154
Wiechert W, Graaf AAd (1997) Bidirectional reaction steps in metabolic networks. Part I: Modeling and simulation of carbon isotope labelling experiments. Biotechnol Bioeng 55:101–117
Sugiyama K, Tagawa S, Toda M (1981) Methods for visual understanding of hierarchical systems. IEEE Trans Syst Man Cybern SMC 11(2):109–125
Acknowledgments
The project was partly funded by the German Ministry BMBF (SysMAP Project), and the German Research Foundation (DFG, project grant WI 1705/13).
Author information
Authors and Affiliations
Corresponding author
Appendix: Implementation details
Appendix: Implementation details
CumoVis (cumomer graph visualization tool) is implemented in Java version 1.5 (http://www.java.sun.com). The three-dimensional visualization of the cumomer network is realized with Java3D (https://java3d.dev.java.net), an OpenGL (http://www.opengl.org) based library for Java. This library requires graphics hardware with full 3D acceleration. CumoVis is available via the authors’ Internet presentation http://www.uni-siegen.de/fb11/simtec/software.
The visualization of an ILN in CumoVis is split into three main tasks (cf. Fig. 8).
Analysis of a carbon transition network
The network generation algorithm introduced in [26] translates the carbon atom transition network given in FluxML (a new file format developed at the University of Siegen in conjunction with 13C flux simulation) into a cumomer network graph with a partition in WLs and a connectivity analysis of the individual WLs of the cumomer network graph reveals their CCs and SCCs. This analysis is not part of the CumoVis software but of an external tool. The obtained information is serialized into a document of another XML language, the CGraph (.cg) format developed for this purpose. It keeps all information about the cumomer network, i.e.
-
all metabolites with all possible cumomers,
-
all reactions with a reference to the involved metabolites,
-
the network partition with all WLs, their reaction edges, their CCs and SCCs and the uplinks between different WLs.
Design of the metabolic network
The underlying metabolic network layout is also designed independently from CumoVis. For that purpose, a flux map editor called Omix can be used. Omix is a follow-up tool of MetVis [47]. It is developed at the University of Siegen in a companion project. For a better lucidity of metabolic networks, Omix—and therefore CumoVis—supports duplicated nodes which are often applied for co-metabolites, like CO2 and H2O.
The Omix output format (Omix Graph Layout .ogl) keeps metabolite and reaction nodes with their coordinates in a two-dimensional layer. Both file formats “ogl” and “cg” are specified in XML. Functionally linked data in aforementioned file formats are associated by the unambiguous names of reactions and metabolites. If no layout data is available, CumoVis can automatically generate a hierarchical layout of the metabolic network by analyzing the topology of the metabolic graph that yields a directed acyclic graph (DAG) of CCs and SCCs. This DAG is designed according Sugiyama’s layered approach [59].
Generating the three-dimensional cumomer graph
Having collected this input data, the CGraph data can be visualized based on the 2D layout in the CumoVis application. Generating the three-dimensional scene of the cumomer network occurs as follows:
-
Arranging the cumomer nodes according the underlying metabolic nodes: The X and Y coordinates of every cumomer result from the underlying metabolite node in level 0 (cf. Fig. 4). Their Z position arises from the position of the appropriate weight level and the position of the single cumomer in that level. In a weight level, all cumomers of the same metabolite are arranged in a pile sorted by their binary number. The height of a weight level arises from the largest number of cumomer nodes of the same metabolite.
-
Generating the edges between the cumomer nodes:
The drawing of edge shapes is quite more difficult than that of nodes, because the positions of their start, control, and end points cannot be taken from the metabolite graph. A reason for this lies on the one hand in the difference between metabolite and cumomer graphs. A metabolite graph is a hyper graph, i.e. every edge can have several sources and destinations whereas edges in a cumomer network have always one source and one destination node (cf. “Cumomer networks” section). On the other hand, a reaction edge in the metabolic network splits into multiple edges in the cumomer network. If the reaction is reversible this number of cumomer edges is even doubled.
To solve this problem of designing edges in the cumomer network automatic layout methods were developed. Every cumomer node sorts its incoming and outgoing edges by the position of the aim node. According to this sort sequence, their start respectively end points are arranged alongside the cumomer. For the shape of the edges a cubic Bézier curve was chosen.
The computation of all edges takes the largest share of the time to generate the 3D network. For example, it needs up to 20 s to build up the citric acid graph, which cannot be avoided.
Certainly, the visualization of full “genome wide” networks makes no sense because they can have more than 100,000 cumomer nodes with stacks of up to 16,000 cumomers. For this reason, the visual approach is limited to medium-sized networks. However, it should be pointed out that MFA is traditionally concerned with the central metabolic pathways, which fulfill this condition.
The graphical user interface contains a selection area that contains the structure of the cumomer graph divided into WLs, CCs, SCCs and single cumomers to control the complexity of the cumomer graph and to choose the path visualization.
The path tracing functionality is solved with a recursive depth search algorithm which searches for a target cumomer and with a maximum depth. In the case of no specified target all possible paths with the given depth are returned. If a cycle is detected, the algorithm breaks the recursion. In order to visualize the aforementioned carbon permutations the sequence of labeled atoms of the source cumomer is permutated according the reactions along the depth search.
Rights and permissions
About this article
Cite this article
Droste, P., Weitzel, M. & Wiechert, W. Visual exploration of isotope labeling networks in 3D. Bioprocess Biosyst Eng 31, 227–239 (2008). https://doi.org/10.1007/s00449-007-0177-1
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00449-007-0177-1