In the era of next-generation sequencing and ubiquitous assembly and binning of metagenomes, new putative genome sequences are being produced from isolate and microbiome samples at ever-increasing rates. Genome-scale metabolic models have enormous utility for supporting the analysis and predictive characterization of these genomes based on sequence data. As a result, tools for rapid automated reconstruction of metabolic models are becoming critically important for supporting the analysis of new genome sequences. Many tools and algorithms have now emerged to support rapid model reconstruction and analysis. Here, we are comparing and contrasting the capabilities and output of a variety of these tools, including ModelSEED, Raven Toolbox, PathwayTools, SuBliMinal Toolbox and merlin.

Many tools have been developed for genome-scale metabolic model (GEM) reconstruction in the past decade [1]. In this review, we focus our efforts on tools for the automated reconstruction of GEMs. Here, we choose to review a mix of early and more recently developed methods, exploring how tool capabilities and outputs have improved over time (Table 1). GEM System [2], AUTOGRAPH [3] and Scrumpy [4] were released in 2006 and were among the first methods attempting to automate GEM reconstruction. AUTOGRAPH uses published models as a template and performs an ortholog search from a target genome to the reference genomes to map genes and their gene–protein reactions. In contrast, GEM System first assigns functions to the genes in a target genome, by conducting homology and orthology searches against the SWISS-PROT and TrEMBL [5] databases. Then, it maps appropriate reactions to metabolic genes, based on EC number matches to the KEGG pathway databases [6]. GEM System also enables users to build models from their own genome annotations based on EC number matching.

Table 1
Comparison between different resources for automated GEM reconstruction
 GEM system merlin Model SEED/KBase Pathway tools Raven toolbox SuBliMinal toolbox 
Input data Annotated or unannotated sequence Annotated or unannotated sequence RAST annotation Annotated sequence Annotated sequence Organisms in KEGG and MetaCyc 
Reference databases KEGG, BioCyc KEGG, TCDB Model SEED database MetaCyc KEGG, Published models KEGG, MetaCyc 
Interface Standalone (GUI) Standalone (GUI) Web Standalone (GUI), Web MATLAB Standalone (cmd line) 
Output SBML SBML SBML, Excel SBML, BioPax SBML, Excel SBML 
Network visualization YES YES YES YES YES NO 
Simulation support NO NO YES YES YES NO 
Integrates gap filling NO NO YES YES YES NO 
 GEM system merlin Model SEED/KBase Pathway tools Raven toolbox SuBliMinal toolbox 
Input data Annotated or unannotated sequence Annotated or unannotated sequence RAST annotation Annotated sequence Annotated sequence Organisms in KEGG and MetaCyc 
Reference databases KEGG, BioCyc KEGG, TCDB Model SEED database MetaCyc KEGG, Published models KEGG, MetaCyc 
Interface Standalone (GUI) Standalone (GUI) Web Standalone (GUI), Web MATLAB Standalone (cmd line) 
Output SBML SBML SBML, Excel SBML, BioPax SBML, Excel SBML 
Network visualization YES YES YES YES YES NO 
Simulation support NO NO YES YES YES NO 
Integrates gap filling NO NO YES YES YES NO 

One significant downside of early methods was that they typically produced nonfunctional models that were incapable of simulating biomass production. Significant additional manual curation was required just to enable biomass production. Scrumpy adopts a software engineering approach where a model is a data object and a programming language (Python) acts as the interface. Scrumpy builds models from the BioCyc database [7] with the option for user input to account for corrections, additions and deletions to the database's annotation. Scrumpy was updated in June 2016 and used to build models for Geobacillus thermoglucosidasius [8] and Solanum lycopersicum L. (tomato) [9].

More recent methods start with a draft reconstruction and provide tools to refine and evaluate the network reconstruction. Figure 1 describes the process of genome-scale metabolic modeling reconstruction. merlin (metabolic models reconstruction using genome-scale information) [10] provides several tools for curation of annotations, along with tools for network reconstruction. As part of its pipeline, merlin implements subcellular localization prediction tools for proteins and metabolites [11,12]. The addition of this feature makes merlin a suitable candidate to build multicompartment models. merlin has been used to build models for both prokaryotic [13] and eukaryotic [14,15] genomes.

GEM reconstruction process.

ModelSEED [16] was the first platform to integrate the capability to generate draft models and perform network refinement and curation, automated gap filling [17] and network evaluation with flux balance analysis and phenotype datasets. ModelSEED exists as a standalone website for model reconstruction (http://modelseed.org), a model reconstruction tool in PATRIC [18] and a more comprehensive set of curation and analysis apps in the DOE Systems Biology Knowledgebase (KBase) [19]. The KBase version of ModelSEED also includes tools for batch model reconstruction, enabling large-scale analyses such as (i) reconstruction of 8,000 models of central metabolism for diverse microbial genomes [20] and (ii) reconstruction of models for 773 members of the human gut microbiome [21]. Pathway Tools [22] stemmed from the development of EcoCyc [23] as a tool to create organism-specific pathway databases. Since then, it has evolved, and its latest release includes a full suite of tools for GEM reconstruction [24].

The RAVEN (Reconstruction, Analysis and Visualization of Metabolic Networks) Toolbox [25] is a MATLAB package that focuses on providing tools for network visualization and analysis in addition to the network reconstruction tools. Similar to merlin, the RAVEN Toolbox also implements tools for prediction of subcellular localization of proteins [11,26]. The SuBliMinal Toolbox [27] offers a distinctive modular approach to the reconstruction process. This modular infrastructure allows users to plug in multiple tools, such as the popular cheminformatics software Marvin Beans (www.chemaxon.com) to determine metabolite charges.

A key point of variation in model reconstruction methods is the input data that they accept (Table 1). GEM System, ModelSEED and merlin allow users to upload a genome sequence file. While GEM System and merlin have annotation tools built directly into their pipelines, ModelSEED links to the RAST annotation system [28,29]. merlin provides an array of tools that allow users to curate and re-annotate the functional annotations in a submitted genome. The RAVEN Toolbox [25] and Pathway Tools require annotated genomes as input. The SuBliMinal Toolbox [27] accepts genomes only from KEGG and MetaCyc [7], limiting the reconstruction process to organisms available on those databases.

Another important area of variation in the model reconstruction tools is their underlying biochemical reference database. All tools compared here use some combination of KEGG, MetaCyc and existing published models as their sources of biochemistry. The ModelSEED internal database absorbs KEGG, MetaCyc and many published models. It is curated to ensure proper charging of compounds and proton balancing of reactions at pH 7. Pathway Tools uses only the MetaCyc database, which is also charged and proton balanced. Both ModelSEED and the RAVEN Toolbox use reactions from published models, as these are usually curated during their respective reconstruction processes. merlin makes additional use of the TCDB transporters database [30] in an effort to better annotate and identify transport reactions. The SuBliMinal Toolbox absorbs the KEGG and MetaCyc databases.

All the model reconstruction tools selected for comparison here offer distinct user interfaces and software distribution practices (Table 1). merlin, Pathway Tools and the SuBliMinal Toolbox are all distributed as standalone applications. The SuBliMinal Toolbox presents a command-line interface, while both merlin and Pathway Tools have graphical user interfaces (GUIs). ModelSEED is available only as a web application. Pathway Tools also has a web interface version. Web interfaces offer the advantage of centralizing all software and reference data in a single location, ensuring that all users immediately benefit from updates to the model reconstruction methods and underlying data. This comes with the disadvantage that some researchers are reluctant to upload their private genome sequences/annotations to the web. The RAVEN Toolbox is distributed as a MATLAB package, having the drawback of requiring a license for MATLAB.

Systems Biology Markup Language (SBML) is the standard model data output for all the tools [31]. ModelSEED and the RAVEN Toolbox also output model data in tabular format. Pathway Tools provides additional output in the BioPax community standard for pathway sharing [32]. Network visualization can be an important feature for both curation and evaluation of models. RAVEN Toolbox uses manually curated CellDesigner [33] maps, while Pathway Tools has its own system for drawing metabolic pathways. merlin and ModelSEED provide a more basic visualization functionality, drawing on top of KEGG maps.

Support for simulations is also essential for model evaluation. FBA is the basic simulation method provided by ModelSEED, Pathway Tools and the RAVEN Toolbox. Tools lacking simulation support require the use of an additional platform, like COBRA Toolbox [34] or OptFlux [35], to apply models for flux simulations.

Integrated gap filling is one of the most important features in a reconstruction, as manual efforts at gap filling can be extremely time-consuming. When performed manually, one has to identify the gaps and candidate reactions to complete a pathway/network. ModelSEED and Pathway Tools provide algorithms that allow for completely automated gap filling. The RAVEN Toolbox approach suggests candidate reactions but requires users to assign the proper gene–protein–reaction association for reactions to be added. merlin provides tools to find gaps, but no support is provided for automated gap filling.

Automated gap-filling solutions still require manual inspection for further refinement, as reactions can be arbitrarily added to restore model connectivity and pathway completeness. This fact calls for the continuous development of improved gap-filling algorithms. To address that issue, all platforms that integrate gap filling usually provide their own algorithms. Those algorithms are variations of the GapFill algorithm, originally purposed by Satish Kumar et al. [17].

One of the drawbacks of GapFill and its variations is the use of mixed-integer linear programing (MILP) to determine the minimum set of reactions to be added to the model. MILP can be difficult to solve, particularly if one lacks commercial optimization software (e.g., CPLEX). In one example case, GapFill was found to take over 14 h to find an optimal solution for a single model of a prokaryote [36]. The biochemistry databases underlying MetaCyc and ModelSEED are now composed of over 10,000 reactions, which further extend the computation time for gap filling. Models with multiple cellular compartments add additional complexity to the gap-filling problem, further extending the computational time. In the development of the heavily compartmentalized human metabolic reconstruction (Recon 1) [37], the authors chose to decompartmentalize the network to facilitate the gap-filling process, although this approach has the disadvantage of potentially coupling reactions that do not co-occur in the same cellular compartment [38].

Recently, methods have been introduced that use linear programing (LP) to substantially reduce the computation time required for gap filling. FASTCORE [39] adopts an LP formulation that was shown to be capable of obtaining good approximations to optimal solutions when compared with a MILP formulation. FastGapFill [38] presents itself as an expansion of FASTCORE that optimizes its LP formulation for heavily compartmentalized organisms. FastGapFill is available as an extension to the COBRA Toolbox [34]. A similar method, FastGapFilling [40] also utilizes an LP formulation; this approach was used to effectively gap-fill Escherichia coli and Saccharomyces cerevisiae networks, performing up to three times faster when compared with a MILP formulation. FastGapFilling was integrated into the simulation framework of MetaFlux [41] that is distributed with Pathway Tools.

Improvements to automated GEM reconstruction tools have substantially accelerated the process of building draft metabolic models from months to a few minutes. Owing to increased accessibility and simple user interfaces, hundreds of curated models are now available in the literature. In industry, GEMs were used for the improved production of bioethanol [42,43], one of biotechnology's most notorious products [44], and for environmental remediation [45]. In research, GEMs have been used in studies to identify new drug targets [46,47], to better understand bacterial evolution [48], to study species interactions within a microbiome [49] and to improve genome functional annotations [50].

With the advent of next-generation sequencing technology, the number of complete genome sequences has been growing exponentially. With the many recent advances in methods for GEM reconstruction, we are now capable of keeping pace to produce draft models for every available sequenced genome. However, much more work is still required. Draft models produced by automated techniques still require substantial curation before they are fully capable of producing reliable qualitative and quantitative predictions. One of the primary causes for the need of manual curation is the inherent errors in the source databases that are passed down to the models. Systematic analysis of the KEGG and MetaCyc databases revealed the existence of unbalanced reactions in both resources [51]. Coincidently, KEGG and/or MetaCyc are used as source databases for all methods described here. Without proper mass and energy balance, models can include inaccurate energy-generation cycles [52] leading to impractical results. The RAVEN Toolbox, merlin and KBase provide tools to flag unbalanced reactions, but no automatic correction is applied. These databases are also not free of incorrect annotations, leading to erroneous assignment of genes to reactions. This is particularly true for nonmodel and newly sequenced organisms that rely on automated annotation pipelines. Ultimately, GEM reconstruction methods must improve further to reduce this required curation for producing a fully predictive model, thus ensuring that modeling can continue to accelerate biological research.

Abbreviations

     
  • FBA

    Flux Balance Analysis

  •  
  • GEM

    genome-scale metabolic model

  •  
  • GPR

    gene–protein–reaction association

  •  
  • GUI

    graphical user interfaces

  •  
  • KBase

    DOE Systems Biology Knowledgebase

  •  
  • LP

    linear programing

  •  
  • MILP

    Mixed-Integer Linear Programing

  •  
  • SBML

    Systems Biology Markup Language

Funding

This work was supported by the U.S. Department of Energy, Office of Biological and Environmental Research; under contract DE-AC02-06CH11357 as a part of the DOE Knowledgebase project.

Competing Interests

The Authors declare that there are no competing interests associated with the manuscript.

References

References
1
Garcia-Albornoz
,
M.A.
and
Nielsen
,
J.
(
2013
)
Application of genome-scale metabolic models in metabolic engineering
.
Ind. Biotechnol.
9
,
203
214
2
Arakawa
,
K.
,
Yamada
,
Y.
,
Shinoda
,
K.
,
Nakayama
,
Y.
and
Tomita
,
M.
(
2006
)
GEM system: automatic prototyping of cell-wide metabolic pathway models from genomes
.
BMC Bioinf.
7
,
168
3
Notebaart
,
R.A.
,
van Enckevort
,
F.H.
,
Francke
,
C.
,
Siezen
,
R.J.
and
Teusink
,
B.
(
2006
)
Accelerating the reconstruction of genome-scale metabolic networks
.
BMC Bioinf.
7
,
296
4
Poolman
,
M.G.
(
2006
)
Scrumpy: metabolic modelling with Python
.
Syst. Biol. (Stevenage)
153
,
375
378
5
Boeckmann
,
B.
,
Bairoch
,
A.
,
Apweiler
,
R.
,
Blatter
,
M.C.
,
Estreicher
,
A.
,
Gasteiger
,
E.
et al. 
(
2003
)
The SWISS-PROT protein knowledgebase and its supplement TrEMBL in 2003
.
Nucleic Acids Res.
31
,
365
370
6
Ogata
,
H.
,
Goto
,
S.
,
Sato
,
K.
,
Fujibuchi
,
W.
,
Bono
,
H.
and
Kanehisa
,
M.
(
1999
)
KEGG: Kyoto Encyclopedia of Genes and Genomes
.
Nucleic Acids Res.
27
,
29
34
7
Caspi
,
R.
,
Foerster
,
H.
,
Fulcher
,
C.A.
,
Kaipa
,
P.
,
Krummenacker
,
M.
,
Latendresse
,
M.
et al. 
(
2008
)
The MetaCyc database of metabolic pathways and enzymes and the BioCyc collection of pathway/genome databases
.
Nucleic Acids Res.
36
,
D623
D6231
8
Ahmad
,
A.
,
Hartman
,
H.B.
,
Krishnakumar
,
S.
,
Fell
,
D.A.
,
Poolman
,
M.G.
and
Srivastava
,
S.
(
2017
)
A genome scale model of Geobacillus thermoglucosidasius (C56-YS93) reveals its biotechnological potential on rice straw hydrolysate
.
J. Biotechnol.
251
,
30
37
9
Yuan
,
H.
,
Cheung
,
C.Y.
,
Poolman
,
M.G.
,
Hilbers
,
P.A.
and
van Riel
,
N.A.
(
2016
)
A genome-scale metabolic network reconstruction of tomato (Solanum lycopersicum L.) and its application to photorespiratory metabolism
.
Plant J.
85
,
289
304
10
Dias
,
O.
,
Rocha
,
M.
,
Ferreira
,
E.C.
and
Rocha
,
I.
(
2015
)
Reconstructing genome-scale metabolic models with merlin
.
Nucleic Acids Res.
43
,
3899
3910
11
Horton
,
P.
,
Park
,
K.J.
,
Obayashi
,
T.
,
Fujita
,
N.
,
Harada
,
H.
,
Adams-Collier
,
C.J.
et al. 
(
2007
)
WoLF PSORT: protein localization predictor
.
Nucleic Acids Res.
35
,
W585
W587
12
Yu
,
N.Y.
,
Wagner
,
J.R.
,
Laird
,
M.R.
,
Melli
,
G.
,
Rey
,
S.
,
Lo
,
R.
et al. 
(
2010
)
PSORTb 3.0: improved protein subcellular localization prediction with refined localization subcategories and predictive capabilities for all prokaryotes
.
Bioinformatics
26
,
1608
1615
13
Resende
,
T.
,
Correia
,
D.M.
,
Rocha
,
M.
and
Rocha
,
I.
(
2013
)
Re-annotation of the genome sequence of Helicobacter pylori 26695
.
J. Integr. Bioinform.
10
,
233
14
Gomes
,
D.
,
Aguiar
,
T.Q.
,
Dias
,
O.
,
Ferreira
,
E.C.
,
Domingues
,
L.
and
Rocha
,
I.
(
2014
)
Genome-wide metabolic re-annotation of Ashbya gossypii: new insights into its metabolism through a comparative analysis with Saccharomyces cerevisiae and Kluyveromyces lactis
.
BMC Genomics
15
,
810
15
Dias
,
O.
,
Pereira
,
R.
,
Gombert
,
A.K.
,
Ferreira
,
E.C.
and
Rocha
,
I.
(
2014
)
iOD907, the first genome-scale metabolic model for the milk yeast Kluyveromyces lactis
.
Biotechnol. J.
9
,
776
790
16
Henry
,
C.S.
,
DeJongh
,
M.
,
Best
,
A.A.
,
Frybarger
,
P.M.
,
Linsay
,
B.
and
Stevens
,
R.L.
(
2010
)
High-throughput generation, optimization and analysis of genome-scale metabolic models
.
Nat. Biotechnol.
28
,
977
982
17
Satish Kumar
,
V.
,
Dasika
,
M.S.
and
Maranas
,
C.D.
(
2007
)
Optimization based automated curation of metabolic reconstructions
.
BMC Bioinf.
8
,
212
18
Wattam
,
A.R.
,
Davis
,
J.J.
,
Assaf
,
R.
,
Boisvert
,
S.
,
Brettin
,
T.
,
Bun
,
C.
et al. 
(
2017
)
Improvements to PATRIC, the all-bacterial bioinformatics database and analysis resource center
.
Nucleic Acids Res.
45
,
D535
D542
19
Arkin
,
A.P.
,
Cottingham
,
R.W.
,
Henry
,
C.S.
,
Harris
,
N.L.
,
Stevens
,
R.L.
,
Maslov
,
S.
et al. 
(
2018
)
KBase: The United States Department of Energy Systems Biology Knowledgebase
.
Nat. Biotechnol.
36
,
566
569
20
Edirisinghe
,
J.N.
,
Weisenhorn
,
P.
,
Conrad
,
N.
,
Xia
,
F.
,
Overbeek
,
R.
,
Stevens
,
R.L.
et al. 
(
2016
)
Modeling central metabolism and energy biosynthesis across microbial life
.
BMC Genomics
17
,
568
21
Magnusdottir
,
S.
,
Heinken
,
A.
,
Kutt
,
L.
,
Ravcheev
,
D.A.
,
Bauer
,
E.
,
Noronha
,
A.
et al. 
(
2017
)
Generation of genome-scale metabolic reconstructions for 773 members of the human gut microbiota
.
Nat. Biotechnol.
35
,
81
89
22
Karp
,
P.D.
,
Paley
,
S.
and
Romero
,
P.
(
2002
)
The Pathway Tools software
.
Bioinformatics
18
(
Suppl 1
),
S225
S232
23
Karp
,
P.D.
,
Riley
,
M.
,
Saier
,
M.
,
Paulsen
,
I.T.
,
Collado-Vides
,
J.
,
Paley
,
S.M.
et al. 
(
2002
)
The EcoCyc database
.
Nucleic Acids Res.
30
,
56
58
24
Karp
,
P.D.
,
Paley
,
S.M.
,
Krummenacker
,
M.
,
Latendresse
,
M.
,
Dale
,
J.M.
,
Lee
,
T.J.
et al. 
(
2010
)
Pathway Tools version 13.0: integrated software for pathway/genome informatics and systems biology
.
Brief. Bioinform.
11
,
40
79
25
Agren
,
R.
,
Liu
,
L.
,
Shoaie
,
S.
,
Vongsangnak
,
W.
,
Nookaew
,
I.
and
Nielsen
,
J.
(
2013
)
The RAVEN toolbox and its use for generating a genome-scale metabolic model for Penicillium chrysogenum
.
PLoS Comput. Biol.
9
,
e1002980
26
Yu
,
C.S.
,
Chen
,
Y.C.
,
Lu
,
C.H.
and
Hwang
,
J.K.
(
2006
)
Prediction of protein subcellular localization
.
Proteins
64
,
643
651
27
Swainston
,
N.
,
Smallbone
,
K.
,
Mendes
,
P.
,
Kell
,
D.
and
Paton
,
N.
(
2011
)
The SuBliMinaL Toolbox: automating steps in the reconstruction of metabolic networks
.
J. Integr. Bioinform.
8
,
186
28
Aziz
,
R.K.
,
Bartels
,
D.
,
Best
,
A.A.
,
DeJongh
,
M.
,
Disz
,
T.
,
Edwards
,
R.A.
et al. 
(
2008
)
The RAST server: rapid annotations using subsystems technology
.
BMC Genomics
9
,
75
29
Overbeek
,
R.
,
Olson
,
R.
,
Pusch
,
G.D.
,
Olsen
,
G.J.
,
Davis
,
J.J.
,
Disz
,
T.
et al. 
(
2014
)
The SEED and the Rapid Annotation of microbial genomes using Subsystems Technology (RAST)
.
Nucleic Acids Res.
42
,
D206
D214
30
Saier
, Jr,
M.H.
,
Tran
,
C.V.
and
Barabote
,
R.D.
(
2006
)
TCDB: the Transporter Classification Database for membrane transport protein analyses and information
.
Nucleic Acids Res.
34
,
D181
D186
31
Hucka
,
M.
,
Finney
,
A.
,
Sauro
,
H.M.
,
Bolouri
,
H.
,
Doyle
,
J.C.
,
Kitano
,
H.
et al. 
(
2003
)
The Systems Biology Markup Language (SBML): a medium for representation and exchange of biochemical network models
.
Bioinformatics
19
,
524
531
32
Demir
,
E.
,
Cary
,
M.P.
,
Paley
,
S.
,
Fukuda
,
K.
,
Lemer
,
C.
,
Vastrik
,
I.
et al. 
(
2010
)
The BioPAX community standard for pathway data sharing
.
Nat. Biotechnol.
28
,
935
942
33
Funahashi
,
A.
,
Matsuoka
,
Y.
,
Jouraku
,
A.
,
Morohashi
,
M.
,
Kikuchi
,
N.
and
Kitano
,
H.
(
2008
)
CellDesigner 3.5: a versatile modeling tool for biochemical networks
.
Proc. IEEE
96
,
1254
1265
34
Schellenberger
,
J.
,
Que
,
R.
,
Fleming
,
R.M.
,
Thiele
,
I.
,
Orth
,
J.D.
,
Feist
,
A.M.
et al. 
(
2011
)
Quantitative prediction of cellular metabolism with constraint-based models: the COBRA Toolbox v2.0
.
Nat. Protoc.
6
,
1290
1307
35
Rocha
,
I.
,
Maia
,
P.
,
Evangelista
,
P.
,
Vilaca
,
P.
,
Soares
,
S.
,
Pinto
,
J.P.
et al. 
(
2010
)
OptFlux: an open-source software platform for in silico metabolic engineering
.
BMC Syst. Biol.
4
,
45
36
Brooks
,
J.P.
,
Burns
,
W.P.
,
Fong
,
S.S.
,
Gowen
,
C.M.
and
Roberts
,
S.B.
(
2012
)
Gap detection for genome-scale constraint-based models
.
Adv. Bioinform.
2012
,
323472
37
Rolfsson
,
O.
,
Palsson
,
B.O.
and
Thiele
,
I.
(
2011
)
The human metabolic reconstruction Recon 1 directs hypotheses of novel human metabolic functions
.
BMC Syst. Biol.
5
,
155
38
Thiele
,
I.
,
Vlassis
,
N.
and
Fleming
,
R.M.
(
2014
)
fastGapFill: efficient gap filling in metabolic networks
.
Bioinformatics
30
,
2529
2531
39
Vlassis
,
N.
,
Pacheco
,
M.P.
and
Sauter
,
T.
(
2014
)
Fast reconstruction of compact context-specific metabolic network models
.
PLoS Comput. Biol.
10
,
e1003424
40
Latendresse
,
M.
(
2014
)
Efficiently gap-filling reaction networks
.
BMC Bioinf.
15
,
225
41
Latendresse
,
M.
,
Krummenacker
,
M.
,
Trupp
,
M.
and
Karp
,
P.D.
(
2012
)
Construction and completion of flux balance models from pathway databases
.
Bioinformatics
28
,
388
396
42
Bro
,
C.
,
Regenberg
,
B.
,
Forster
,
J.
and
Nielsen
,
J.
(
2006
)
In silico aided metabolic engineering of Saccharomyces cerevisiae for improved bioethanol production
.
Metab. Eng.
8
,
102
111
43
Ng
,
C.Y.
,
Jung
,
M.Y.
,
Lee
,
J.
and
Oh
,
M.K.
(
2012
)
Production of 2,3-butanediol in Saccharomyces cerevisiae by in silico aided metabolic engineering
.
Microb. Cell Fact.
11
,
68
44
Otero
,
J.M.
,
Panagiotou
,
G.
and
Olsson
,
L.
(
2007
)
Fueling industrial biotechnology growth with bioethanol
.
Adv. Biochem. Eng. Biotechnol.
108
,
1
40
45
Singh
,
J.S.
,
Abhilash
,
P.C.
,
Singh
,
H.B.
,
Singh
,
R.P.
and
Singh
,
D.P.
(
2011
)
Genetically engineered bacteria: an emerging tool for environmental remediation and future research perspectives
.
Gene
480
,
1
9
46
Chavali
,
A.K.
,
D'Auria
,
K.M.
,
Hewlett
,
E.L.
,
Pearson
,
R.D.
and
Papin
,
J.A.
(
2012
)
A metabolic network approach for the identification and prioritization of antimicrobial drug targets
.
Trends Microbiol.
20
,
113
123
47
Kim
,
H.U.
,
Kim
,
S.Y.
,
Jeong
,
H.
,
Kim
,
T.Y.
,
Kim
,
J.J.
,
Choy
,
H.E.
et al. 
(
2011
)
Integrative genome-scale metabolic analysis of Vibrio vulnificus for drug targeting and discovery
.
Mol. Syst. Biol.
7
,
460
48
Pál
,
C.
,
Papp
,
B.
,
Lercher
,
M.J.
,
Csermely
,
P.
,
Oliver
,
S.G.
and
Hurst
,
L.D.
(
2006
)
Chance and necessity in the evolution of minimal metabolic networks
.
Nature
440
,
667
670
49
Henry
,
C.S.
,
Bernstein
,
H.C.
,
Weisenhorn
,
P.
,
Taylor
,
R.C.
,
Lee
,
J.Y.
,
Zucker
,
J.
et al. 
(
2016
)
Microbial community metabolic modeling: a community data-driven network reconstruction
.
J. Cell. Physiol.
231
,
2339
2345
50
Reed
,
J.L.
,
Patel
,
T.R.
,
Chen
,
K.H.
,
Joyce
,
A.R.
,
Applebee
,
M.K.
,
Herring
,
C.D.
et al. 
(
2006
)
Systems approach to refining genome annotation
.
Proc. Natl Acad. Sci. U.S.A.
103
,
17480
17484
51
Altman
,
T.
,
Travers
,
M.
,
Kothari
,
A.
,
Caspi
,
R.
and
Karp
,
P.D.
(
2013
)
A systematic comparison of the MetaCyc and KEGG pathway databases
.
BMC Bioinf.
14
,
112
52
Fritzemeier
,
C.J.
,
Hartleb
,
D.
,
Szappanos
,
B.
,
Papp
,
B.
and
Lercher
,
M.J.
(
2017
)
Erroneous energy-generating cycles in published genome scale metabolic networks: identification and removal
.
PLoS Comput. Biol.
13
,
e1005494