Biopax to sbml converter

The need biological and biochemical processes across all layers and vari- to combine both formats to use the knowledge from a multitude of ous levels of detail. The Biological Pathway Exchange Language databases in various applications becomes more and more urgent. We demonstrate its functionality by rage of in-silico models. Published by Oxford University Press. The most important classes are spe- cies, describing reactive species, and reactions, which interconnect Gene SBO informational molecule segment species elements.

The SBML core specifi- Protein SBO polypeptide chain cation provides several constructs to describe quantitative processes, such as Dna SBO deoxyribonucleic acid events, rules, constraints, reactions, etc. SBML does not contain specific entities that can be Species exhibit discrete states, representing their activities that are chan- derived from an SBML species. The common way to separate different genomic entities ged using transitions.

This table specifies the SBO elements. The sign attribute of the input elements describes whether the terms that we used to distinguish between various cellular entities in SBML. Dual means that the transition can operate both activa- ting positive and inhibiting negative. In contrast, unknown is assigned to the input if the transition effect is not further specified.

If, in a qualita- 2. Firstly, the pathway organism tive model, the activity of protein A inhibits the activity of protein B, this is determined by searching for the BioSource reference in the BioPAX would be represented as a transition with an input A, whose sign attribute file. Both models correspond to the complete pathway represented in the BioPAX file. There is one superclass called for each PhysicalEntity. Two main classes Entity, i. Physical- the species is annotated with the corresponding SBO term Courtot Entity describes molecules, such as proteins, complexes, small molecules, et al.

The used SBO terms are listed in Table 1. Interaction is split into Con- tion of the PhysicalEntity. The default compartment is set if the trol and Conversion, which can be separated in several subclasses see CellularLocation is not known. Figure 1. These identifers Level 1 is exclusively able to describe metabolic interactions, whereas are unique and facilitate the automated annotation of this species descri- Level 2 supports signaling pathways and molecular interactions.

In addition bed in the fourth step. If there exists no Gene ID but a gene symbol, the to Level 2, gene-regulatory networks and genetic interactions can be descri- gene symbol is mapped to a Gene ID.

Level 3 is not downwards com- 2. An SBML transition describes relationships ses in lower case typewriter font and the specification of Level 3 denotes between molecules that cannot be translated into reactions. Examples them in upper case typewriter font. For better readability of this paper, all for such relationships are enzyme-enzyme relations, protein-protein inter- BioPAX element names begin with capital letters and refer to Level 2 and 3.

The translation of the Conversion elements is straightforward, because 2. Furthermore, the Models extension qual.

Schaefer et al. The translation of the BioPAX Level 2 and Level 3 The translation of Control elements is more complicated, because they pathway files is performed in four steps: 1 initializing the models, 2 trans- are translated into a transition or a reaction depending on enclo- lation of PhysicalEntity elements, 3 translation of Interaction sed Control elements.

Control elements always consist of zero or more elements, and 4 annotation of all species. An overview of the mapping Controller and zero or one Controlled elements.

The dashed rectangles denote elements, which are only available in Level 3. All other elements occur in both levels. Lines, ending with a diamond, indicate elements that are contained in other elements. This translation dependency is visualized with black dashed lines. A detailed translation description of those elements is shown in Table 2. To obtain For nearly all Control elements a ControlType is assigned describing identifiers for those databases, we map the Entrez Gene identifier, which the relationship between the enclosed elements i.

The goal of those annotations is to provide models whose components 2. Finally, the can uniquely be identified by any application and be linked to external data SBML instances are further annotated. The BioPAX specification allows sources. Hence, former conversion approaches from BioPAX to SBML did either incorrectly convert those relations to reactions or simply removed them during the translation.

To fill this gap, the SBML community has recently developed the qual specification, which allows users to model arbitrary transitions between species. Furthermore, the models themselves just provide the base for further analysis or visualization methods.

Other applications, such as Clandestine Funahashi et al. Therefore, most of those applications have certain requirements on the models. For example, to uniquely map mass spectrometry data on a model, it may be required for the model to have UniProt IDs.

To match mRNA expression data or perform gene set enrichment analyses, Entrez Gene identifiers might be required. Consequently, we provide all annotations that we could gather from the input BioPAX files also in the SBML files and further annotate all species with a plethora of additional identifiers.

The qual extension has been created recently and, thus, might not be supported by all applications, yet. Therefore, we decided to build joint SBML core and qual models. These models are compatible with older applications that do not yet support qual but still can read all species and reactions. Newer applications that are ready to handle relations can read the additional qual model and process all information that was also available in the BioPAX file.

The reason for converting both levels was the additional description possibility of gene-regulatory networks and genetic interactions in BioPAX Level 3, which is not supported by Level 2 pathway models. Since older simulation applications still work with BioPAX Level 2, we also translated these files into SBML in order to prevent loss of information and to be able to use these models, too.

Only a few approaches exist to convert BioPAX to SBML and the existing ones use a simple one-to-one conversion without augmenting the file content for further modeling steps.

Sybill is a stand-alone tool that is also integrated in the quantitative modeling environment VCell Slepchenko et al. Table 3 compares these programs based on defined criteria. Sybill converts BioPAX Level 2 and Level 3 files and has a very comfortable graphical user interface allowing the user to manipulate the conversion result.

Unfortunately, the converted SBML files are not complete and the validator from sbml. Additionally, some groups and pathway links are missing, too. BiNoM generates a complete conversion result, but the validator also reports errors due to empty listOf elements and due to the wrong order of these elements. Another feature of BiNoM is that it can separately visualize reaction networks, pathway structure and protein—protein interaction networks out of one BioPAX file.

All approaches avoid the translation of duplicate species. Conversion between different formats is important in all parts of computer science. In many cases, conversion leads to errors or a loss of information. But with SBML Level 3 Version 1 and the addition of extensions to the specifications, in particular the qualitative models extension qual , it is now possible to create accurate and specification-conform SBML code.

Using this extension, we produced error-free SBML models while minimizing or even eliminating the loss of information during the translation. All relations from the BioPAX documents that could not be converted to exact reactions have been included as qualitative transitions between qualitative species.

These models can easily be used, e. A conversion is valid if the validator from sbml. The No duplicate entities criterion is important for modeling purposes to guarantee that a species is only mentioned once.

A converter is robust if it can handle all tested files from the Pathway Interaction Database and is able to convert a BioPAX file, which contains no Pathway element. Finally, the provenance criterion denotes if the file history and conversion tool information is saved in the converted SBML file.

Google Scholar. Oxford University Press is a department of the University of Oxford. It furthers the University's objective of excellence in research, scholarship, and education by publishing worldwide. Sign In or Create an Account. Sign In. Advanced Search. Search Menu. Article Navigation.

Close mobile search navigation Article Navigation. Volume Article Contents Abstract. Oxford Academic. Clemens Wrzodek. Florian Mittag. Johannes Eichner. Nicolas Rodriguez. Andreas Zell. Associate Editor: Martin Bishop. Revision received:. Select Format Select format. Permissions Icon Permissions. Abstract Motivation: The biological pathway exchange language BioPAX and the systems biology markup language SBML belong to the most popular modeling and data exchange languages in systems biology.

The validation report for the converted model is pretty good and include only a single type of error due to the lack of annotations to some entities in the SBML model. The outstanding error with the report is related to EntityReference instances that don't have any UnificationXref s associated with them.

This is not an artifact of the conversion, but rather a result of the lack of annotations in the Recon 2 model, where some of the SmallMolecule species do not have any annotations to them, hence don't have any UnificationXref s.

Skip to content. Star 0. Branches Tags. Could not load branches. Could not load tags. Latest commit.