Table 3

Comparison of different available converters for KEGG pathways
KEGG2SBML BN++ KEGGconverter KGML2BioPAX KGML2SBML KEGGtranslator
Version 1.5.0 1.1 n/a n/a 1.2 2.0
Release date 2008-07-28 2009-04-22 2009-12-18 2010-06-03 2011-07-04 2012-06-04
Authors Funahashiet al. Küntzeret al. Moutseloset al. Leeet al. Wrzodeket al.
Supported model formats
SBML
BioPAX - - -
Generic translation features
Machine interpretable
Human interpretable - -
Signaling pathways - - - - -
Complete - -
No duplicate entries -
No duplicate reactions -
Unbundle reactions - - - -
Revision of reactions - - -
Stoichiometry - - - - -
SBML
Valid n/a -
Level.Version 1.1 up to 2.3 n/a 2.1 2.4 2.4 2.4, 3.1
SBO terms - n/a - -
Notes - n/a - -
Annotations - n/a - -
BioPAX
Valid n/a - n/a - n/a
Level n/a 2 n/a 2 n/a 2, 3
Appropriate classes n/a n/a - n/a
Notes n/a - n/a - n/a
Annotations n/a n/a - n/a
SM annotations n/a - n/a - n/a

This table compares various applications that can convert KEGG pathways to BioPAX or SBML models. A checkmark (✓) is given, if the corresponding converter completely fulfills all requirements, a circle (∘) states that the requirements are only met partially or incorrectly and a minus (-) indicates features, which are not contained at all. ‘n/a’ indicates that a criterion is not applicable to a converter. A model is Machine interpretable if entities in the model can directly be mapped to a database. The criterion Human interpretable indicates that a model somehow assigns human readable names or gene symbols to entities. Signaling pathways are supported if the converters can read and convert KEGG models with relations. A conversion is complete if every relevant reaction of a KGML pathway also occurs in any form in the translated document. For visualization purposes, KGML files often contain multiple copies of entries or reactions. These duplicates should be removed. The contained reactions are often bundled (multiple reactions are summarized as one) or miss some reaction participants. Revision of reactions refers to the completion of missing reaction participants. The stoichiometry is not contained in KGML documents and must be parsed from reaction equations in the KEGG REACTION database. To test the validity of the models, we used the corresponding validators from SBML.org and BioPAX.org. A model is marked as valid, if the validator does not return any errors. For SBML, we further inspect if the models contain SBO terms. It is further recommended to include notes, such as human readable descriptions, and annotations (e.g., cross-references in form of CV terms, MIRIAM URNs, Xrefs). Only for BioPAX, it is important to use the appropriate classes (instances of smallMolecule for small molecules and instances of protein for proteins) and a nice feature to fill the available BioPAX fields for chemical formula or molecular weight of small molecules (SM annotations).

Wrzodek et al.

Wrzodek et al. BMC Systems Biology 2013 7:15   doi:10.1186/1752-0509-7-15

Open Data