RAVEN 2.0: A versatile toolbox for metabolic network reconstruction and a case study on Streptomyces coelicolor.
Wang H, Marcišauskas S, Sánchez BJ, Domenzain I, Hermansson D, Agren R, Nielsen J, Kerkhoven EJ. · 2018
Read RAVEN 2.0: A versatile toolbox for metabolic network reconstruction and a case study on Streptomyces coelicolor. by Wang H, Marcišauskas S, Sánchez BJ, Domenzain I, Hermansson D, Agren R, Nielsen J, Kerkhoven EJ. (2018) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc6207324.txt.
Introduction
Wang and co-authors present RAVEN 2.0, a software toolbox. The toolbox builds genome-scale models of cell metabolism. Such models list all chemical reactions in a cell. They help predict how cells react to change. The work appeared in PLoS Computational Biology. It includes a case study on a soil bacterium.
The new version can build models from scratch. It uses the MetaCyc database for that task. It also has a rebuilt KEGG-based pipeline. The team built fresh models for Streptomyces coelicolor. That bacterium makes antibiotics. Their models matched most of a hand-built gold standard. They then used the fresh model to improve that standard. The result, Sco4, is the fullest model yet for this species.
Catalog record
| CREATOR | Wang H, Marcišauskas S, Sánchez BJ, Domenzain I, Hermansson D, Agren R, Nielsen J, Kerkhoven EJ. |
|---|---|
| YEAR | 2018 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 46 KB |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
RAVEN 2.0: A versatile toolbox for metabolic network reconstruction and a case study on Streptomyces coelicolor By Hao Wang, Simonas Marcišauskas, Benjamín J. Sánchez, Iván Domenzain, Daniel Hermansson, Rasmus Agren, Jens Nielsen, Eduard J. Kerkhoven, Christos A. Ouzounis Published in: PLoS Computational Biology DOI: 10.1371/journal.pcbi.1006541 ABSTRACT RAVEN is a commonly used MATLAB toolbox for genome-scale metabolic model (GEM) reconstruction, curation and constraint-based modelling and simulation. Here we present RAVEN Toolbox 2.0 with major enhancements, including: (i) de novo reconstruction of GEMs based on the MetaCyc pathway database; (ii) a redesigned KEGG-based reconstruction pipeline; (iii) convergence of reconstructions from various sources; (iv) improved performance, usability, and compatibility with the COBRA Toolbox. Capabilities of RAVEN 2.0 are here illustrated through de novo reconstruction of GEMs for the antibiotic-producing bacterium Streptomyces coelicolor. Comparison of the automated de novo reconstructions with the iMK1208 model, a previously published high-quality S. coelicolor GEM, exemplifies that RAVEN 2.0 can capture most of the manually curated model. The generated de novo reconstruction is subsequently used to curate iMK1208 resulting in Sco4, the most comprehensive GEM of S. coelicolor, with increased coverage of both primary and secondary metabolism. This increased coverage allows the use of Sco4 to predict novel genome editing targets for optimized secondary metabolites production. As such, we demonstrate that RAVEN 2.0 can be used not only for de novo GEM reconstruction, but also for curating existing models based on up-to-date databases. Both RAVEN 2.0 and Sco4 are distributed through GitHub to facilitate usage and further development by the community (https://github.com/SysBioChalmers/RAVEN and https://github.com/SysBioChalmers/Streptomyces_coelicolor-GEM). INTRODUCTION Genome-scale metabolic models (GEMs) are comprehensive in silico representations of the complete set of metabolic reactions that take place in a cell [1]. GEMs can be used to understand and predict how organisms react to variations on genetic and environmental parameters [2]. Recent studies demonstrated the extensive applications of GEMs in discovering novel metabolic engineering strategies [3]; studying microbial communities [4]; finding biomarkers for human diseases and personalized and precision medicines [5,6]; and improving antibiotic production [7]. With the increasing ease of obtaining whole-genome sequences, significant challenges remain to translate this knowledge to high-quality GEMs [8]. To meet the increasing demand of metabolic network modelling, the original RAVEN (Reconstruction, Analysis and Visualization of Metabolic Networks) toolbox was developed to facilitate GEM reconstruction, curation, and simulation [9]. In addition to facilitating the analysis and visualization of existing GEMs, RAVEN particularly aimed to assist semi-automated draft model reconstruction, utilizing existing template GEMs and the KEGG database [10]. Since publication, RAVEN has been used in GEMs reconstruction for a wide variety of organisms, ranging from bacteria [11], archaea [12] to human gut microbiome [13], eukaryotic microalgae [14], parasites [15–17], and fungi [18], as well as various human tissues [19,20] and generic mammalian models with complex metabolism [21,22]. As such, the RAVEN toolbox has functioned as one of the two major MATLAB-based packages for constraint-based metabolic modelling, together with the COBRA Toolbox [23–25]. Here, we present RAVEN 2.0 with greatly enhanced reconstruction capabilities, together with additional new features (Fig 1, Table 1). A prominent enhancement of RAVEN 2.0 is the use of the MetaCyc database in assisting draft model reconstruction. MetaCyc is a pathway database that collects only experimentally verified pathways with curated reversibility information and mass-balanced reactions [26]. RAVEN 2.0 can leverage this high-quality database to enhance the GEM reconstruction process. While the functionality of the original RAVEN toolbox was illustrated by reconstructing a GEM of Penicillium chrysogenum [9], we here demonstrate the new and improved capabilities and wide applicability of RAVEN 2.0 through reconstruction of a GEM for Streptomyces coelicolor. [Figure: Schematic overview of RAVEN toolbox version 2.0.RAVEN was significantly enhanced for de novo draft model reconstruction by integrating knowledge from different sources (e.g. KEGG and MetaCyc). It has better import/export support to relevant formats, and especially improved compatibility with the COBRA Toolbox by resolving previously conflicting function names and providing a bi-directional model conversion function. As a MATLAB toolbox, RAVEN provides one unified environment for both model reconstruction and simulation analysis (e.g. FSEOF) and allows scripting for more flexible operations.] [Table: Feature comparison of GEM reconstruction toolboxes.] S. coelicolor is a representative species of soil-dwelling, filamentous and Gram-positive actinobacterium harbouring enriched secondary metabolite biosynthesis gene clusters [27,28]. As a well-known pharmaceutical and bioactive compound producer, S. coelicolor has been exploited for antibiotic and secondary metabolite production [29]. The first published GEM for S. coelicolor, iIB711 [30], was improved through an iterative process resulting in the GEMs iMA789 [31] and iMK1208 [32]. The most recent GEM, iMK1208, is a high-quality model that includes 1208 genes and 1643 reactions and was successfully used to predict metabolic engineering targets for increased production of actinorhodin [32]. Here, we demonstrate how the new functions of RAVEN can be used for de novo reconstruction of a S. coelicolor GEM, using comparison to the existing high-quality model iMK1208 as benchmark. The use of three distinct de novo reconstruction approaches enabled capturing most of the existing model, while complementary reactions found through the de novo reconstructions gave the opportunity to improve the existing model. After manual curation, we included 402 new reactions into the GEM, with 320 newly associated enzyme-coding genes, including a variety of biosynthetic pathways for known secondary metabolites (e.g. 2-methylisoborneol, albaflavenone, desferrioxamine, geosmin, hopanoid and flaviolin dimer). The updated S. coelicolor GEM is released as Sco4, which can be used as an upgraded platform for future systems biology research on S. coelicolor and related species. RESULTS Genome-scale reconstruction and curation with RAVEN 2.0 RAVEN 2.0 aims to provide a versatile and efficient toolbox for metabolic network reconstruction and curation (Fig 1). In comparison to other solutions for GEM reconstruction (Table 1), the strength of RAVEN is its ability of semi-automated reconstruction based on published models, KEGG and MetaCyc databases, integrating knowledge from diverse sources. A brief overview of RAVEN capabilities is given here, while more technical details are stated in Material & Methods, and detailed documentation is provided for individual functions in the RAVEN package. RAVEN supports two distinct approaches to initiate GEM reconstruction for an organism of interest: (i) based on protein homology to an existing template model, or (ii) de novo using reaction databases. The first approach requires a high-quality GEM of a phylogenetically closely related organism, and the functions getBlast and getModelFromHomology are used to infer homology using bidirectional BLASTP and build a subsequent draft model. Alternatively, de novo reconstruction can be based on two databases: KEGG and MetaCyc. For KEGG-based reconstruction, the user can deploy getKEGGModelForOrganism to either rely on KEGG-supplied annotations—KEGG currently includes over 5000 genomes—or query its protein sequences for similarity to HMMs that are trained on genes annotated in KEGG. MetaCyc-based reconstruction can be initiated with getMetaCycModelForOrganism that queries protein sequences with BLASTP for homology to enzymes curated in MetaCyc, while addSpontaneous retrieves relevant non-enzyme associated reactions from MetaCyc. Regardless of which (combination of) approach(es) is followed, a draft model is obtained that requires further curation to result in a high-quality reconstruction suitable for simulating flux distributions. Various RAVEN functions aid in this process, including gapReport that runs a gap analysis and reports e.g. dead-end reactions and unconnected subnetworks that indicate missing reactions and gaps in the model, in addition to reporting