kegg2bipartitegraph 0.2.3

Reconstruct metabolic bipartite graph using KEGG

kegg2bipartitegraph

kegg2bipartitegraph is a Python package to create KEGG graphs. The main idea of this package is to create metabolic graphs from KEGG database according to the ones used in the article Weber Zendrera et al. (2021). In this article, the authors creates the metabolic networks from the organism of KEGG (accessible in this github repository). Using annotation (from EsMeCaTa, eggnog-mapper, KofamKOALA or GenBank files) or a KEGG organism ID, kegg2bipartitegraph maps the EC (and Gene Ontology Terms) to KEGG reactions and reconstruct metabolic networks associated with the organism following the proposal of the article Weber Zendrera et al. (2021).

Installation

This package can be installed using pip:

pip install kegg2bipartitegraph

Usage

Kegg2bipartitegraph can be called by using the command k2bg. It is divided in different parts:

• k2bg reference an optional ones that creates the reference data (especially the universal reference metabolic graphs). By default, these data are precomputed and available within the package located in kegg2bipartitegraph/data/kegg_model. It is not meant to be used by the user as it takes a long time to process, only used the precomputed reference files or archived files.

• subcommands to reconstruct metabolic graphs from different inputs:

  • k2bg reconstruct_from_esmecata takes as input the annotation output folder from EsMeCaTa and reconstruct the metabolic networks associated with each taxon.
  • k2bg reconstruct_from_eggnog takes as input the annotation output file from eggnog-mapper to map the EC to KEGG reactions.
  • k2bg reconstruct_from_kofamkoala takes as input the result from KofamKOALA.
  • k2bg reconstruct_from_picrust takes as input the result folder from picrust2.
  • k2bg reconstruct_from_organism takes as input an organism ID from KEGG (such as hsa for human or eco for Escherichia coli). You can find the list of the accessile organisms in KEGG website.
  • k2bg reconstruct_from_genbank takes as input a folder containing GenBank files.

• k2bg scope subcommand to compute scope according to the procedure presented in the article of Weber Zendrera et al. (2021). It expects as input a folder containing graphml files. By default, it uses the four seed files from the article of Weber Zendrera et al. (2021) but you can provide your own seed file as a text file containing KEGG metabolites. It computes the scope and returns in a json file the activated reactions and the accessible metabolites for each graphml file of the input folder.

Online / Offline requirements

Multiple subcommands can be used to reconstruct draft networks. Some of them required an internet connection to work, you can see which ones in the following table:

Subcommands	Online	Offline
reconstruct_from_esmecata	(Mapping of KOs)	X (without mapping KOs)
reconstruct_from_eggnog		X
reconstruct_from_kofamkoala		X
reconstruct_from_picrust		X
reconstruct_from_genbank		X
reconstruct_from_organism	X
reference	X

Reference model

The k2bg reference is to be used only if you want to update the KEGG reference data. First, delete the data contain in kegg2bipartitegraph/data/kegg_model, then use this command to download all the required data. This step is long, it is advised to not use it.

It will create several files:

• kegg_model.sbml: a universal graph containing most of the reactions contained in KEGG database. The stoechiometry is simplified as these metabolic networks are created in order to be used in topological analysis, these changed can be looked at in file kegg_removed_changed_reaction.tsv. So they are not supposed to be used with other methods (such as Constraint-Based Modelling).

• kegg_model.graphml: the metabolic bipartite graph associated with the kegg_model.sbml file.

• several mapping files to go from annotation (especially EC number) to KEGG reactions: kegg_compound_name.tsv, kegg_mapping.tsv and kegg_pathways.tsv. Also a file to use KEGG hierarchy for pathway/module/metabolite: kegg_hierarchy.json. A mapping file ec_to_gos.tsv to convert Gene Ontology terms to EC number using the go2ec file provided by the Gene Ontology Consortium.

• kegg_metadata.json: a metadata file showing the metadata for the creation of the reference files for kegg2bipartitegraph.

Output files of reconstruction command

The reconstruction subcommands will reconstruct draft metabolic networks by mapping the annotation with the metabolic graphs contained in kegg2bipartitegraph. Such as in the graph made by Weber Zendrera et al. (2021), 14 cofactors are removed (H2O, ATP, ADP, NAD+, NADH, NADP+, NADPH, CO2, ammonia, sulfate, thioredoxin, phosphate, pyrophosphate (PPi), and H+) when creating these networks.

Then it will create multiple files:

• a sbml file containing the metabolic network that can be used with topological analysis methods (such as MeneTools, MiSCoTo or Metage2Metabo).

• a graphml file containing the metabolic network as a bipartite graph. At this moment, it is not used, but I am currently adaptating the scope method of Weber Zendrera et al. (2021) to automatise its use with this package.

• tsv files indicating the pathways/modules contained in the metabolic networks, their completness ratio and the associated reactions.

• a tsv file showing KO information if the option has been used.

• a module_class.tsv showing the absence/presence of generic module classes in the organism.

• several statistics/metadata/log files.

Output files of scope command

For each seed file, an output subfolder is created in the output fodler and containes a accessibility.json file. This json file contains two keys:

• producible_metabolites indicates for each input graphml file the accessible metabolites.

• activated_reactions indicates for each input graphml file the activated reactions.

Right now, the metabolic networks created by kegg2bipartitegraph do not reproduce exactly the ones from the article of Weber Zendrera et al. (2021). This leads to a bigger scope.

Citation

At this moment, there are no articles for kegg2bipartitegraph, if you use it and want to cite it, you can cite this GitHub.

Also, please cite the following article:

• the article made by Adèle Weber Zendrera et al. (2021) that proposed this method:

Weber Zendrera, A., Sokolovska, N. & Soula, H.A. Functional prediction of environmental variables using metabolic networks. Scientific Reports 11, 12192 (2021). https://doi.org/10.1038/s41598-021-91486-8

• the KEGG database:

Kanehisa, M., Furumichi, M., Sato, Y., Kawashima, M., Ishiguro-Watanabe, M. KEGG for taxonomy-based analysis of pathways and genomes, Nucleic Acids Research, Volume 51, Issue D1, Pages D587–D592 (2023). https://doi.org/10.1093/nar/gkac963

Kanehisa, M., Goto, S. KEGG: Kyoto Encyclopedia of Genes and Genomes, Nucleic Acids Research, Volume 28, Issue 1, Pages 27–30 (2000). https://doi.org/10.1093/nar/28.1.27

Kanehisa, M. Toward understanding the origin and evolution of cellular organisms. Protein Science. 28: 1947–1951 (2019). https://doi.org/10.1002/pro.3715

• bioservices for the query on KEGG:

Cokelaer, T., Pultz, D., Harder, L., M., Serra-Musach, J., Saez-Rodriguez, J., BioServices: a common Python package to access biological Web Services programmatically, Bioinformatics, Volume 29, Issue 24, Pages 3241–3242 (2013). https://doi.org/10.1093/bioinformatics/btt547

• libsbml for the handling of the SBML:

Bornstein B. J., Keating S. M., Jouraku, A., Hucka, M., LibSBML: an API Library for SBML, Bioinformatics, Volume 24, Issue 6, Pages 880–881 (2008). https://doi.org/10.1093/bioinformatics/btn051

• networkx for the creation of the graphml:

Hagberg A. A., Schult D. A., Swart P. J. Exploring Network Structure, Dynamics, and Function using NetworkX, in: Varoquaux, G., Vaught, T., Millman, J. (Eds.), . Presented at the Proceedings of the Python in Science Conference (SciPy) 2008. 11–15. http://conference.scipy.org/proceedings/SciPy2008/paper_2/

If you have used the subcommand reconstruct_from_genbank, please also cite:

• Biopython for GenBank parsing:

Cock, P.J.A., Antao, T., Chang, J.T., Chapman, B.A., Cox, C.J., Dalke, A., Friedberg, I., Hamelryck, T., Kauff, F., Wilczynski, B., de Hoon, M.J.L. Biopython: freely available Python tools for computational molecular biology and bioinformatics. Bioinformatics 2009, 25, 1422–1423 https://doi.org/10.1093/bioinformatics/btp163.