BioModels Database logo

BioModels Database


BioModels Database Annotation Information

This document contains a general introduction to the annotation of models performed by the curators of BioModels Database.

What is annotation?

The process of annotating a model consists in identifying the model components. This is achieved by linking model components with terms from controlled vocabularies and entries in data resources. This is similar to attaching cross-references to the model with the addition of qualifiers, in order to explicitly define the relationship between the model component and the target element.

Why annotating models?

In publications describing models, the names of different model elements such as genes, proteins and metabolites, or the organisms from which the model is derived, are often denoted by biologically non-meaningful names. The entire paper describing the model should be read to know the relationships between each model element and their biological meaning. To make this easier and to allow users to directly relate the model as well as the model elements to the relevant biological concepts, efforts are being put to enrich models with annotations.

Annotated models provide numerous advantages. Their understanding by both users and software tools is highly enhanced, which therefore lower the barriers of their reuse. Models can more accurately be searched in a repository. Tasks such as model comparison, clustering of similar models and integration with other models are made possible with annotated models. Annotation also facilitates the conversion of models to other formats.

How to encode annotation?

Annotation in the SBML model files provided by BioModels Database is encoded using the SBML controlled annotation scheme. Please refer to the SBML specifications (specially the section entitled "A standard format for the annotation element") for more detailed information. This scheme makes use of MIRIAM URIs (Juty et al. 2012), qualifiers, but also Dublin Core, vCard and RDF.

Example of annotation in SBML
Figure 1 Example of one annotation block from an SBML file.

The Figure 1 shows that the SBML species L_EGFR is actually a complex (usage of the qualifier hasPart) composed of two proteins (presence of 2 URIs pointing to UniProt nested under the qualifier). This was shown for information purposes only, as many tools allow the creation and edition of annotation without the need for the user to understand how the information is ultimately stored in the file.

Details about MIRIAM URIs can be found from the Registry website (which provides services to generate those URIs).

How to identify relevant resources for annotation?

The choice of which resource to use depends on the kind of model component which needs to be annotated.

To date, model elements in BioModels Database are annotated using more than 45 different external resources. Some of the predominantly used external resources for model annotations in BioModels Database are:

Qualifiers in the annotation

The qualifiers in the annotation of models, represent the relationship between the model element and its annotation. The list of all existing qualifiers, their definition, and how they can be used within a SBML model file, can be found on the website.

How are annotations created and edited?

The search for the biological entities which are represented by the model components is performed manually. The information is coming from the model, the associated publication, and any notes or correspondence from the modeller(s). Once the relevant biological entities have been found, their formal identification is obtained from key word searches in the appropriate external resources.

Once the relevant biological entities have been identified, several tools are available to store this information in the SBML model files. A few of them are listed below, but this is far from exhaustive. Please refer to the SBML Software Guide for a more complete list.

BioModels Database provides an annotation interface (Figure 2). Access to this feature is currently restricted to its team of curators. The appropriate qualifiers and external data resource names are selected from a pull down menu. The entity identifier has to be typed in. The interface makes use of Ajax technologies in order to provide a dynamical way to annotate large models. The system automatically and transparently creates the annotation blocks (including the qualifiers and MIRIAM URIs, of the form shown in Figure 1) and includes it in the SBML file.

Annotation interface of BioModels Database
Figure 2 Annotation Interface of BioModels Database.

Model annotation can also be done using SBMLeditor (Figure 3). Similarly to BioModels Database's annotation interface, the software takes care of generating the whole annotation blocks based on the information the user entered in the provided fields.

Annotation interface of SBMLeditor
Figure 3 Annotation interface of SBMLeditor.

COPASI also provides some annotation features (Figure 4). It allows users to create and edit all possible annotation. Further information can be found from the COPASI documentation.

Annotation interface of COPASI
Figure 4 Annotation interface of COPASI.

CellDesigner provides features where notes and MIRIAM annotations can be added and edited (Figure 5). Further information can be found from the CellDesigner documentation.

Annotation interface of CellDesigner
Figure 5 Annotation interface of CellDesigner.

semanticSBML provides a very easy-to-use annotation interface. The interface includes a search feature to find appropriate resource identifiers to annotate model elements (Figure 6).

Annotation interface of semanticSBML
Figure 6 Annotation interface of semanticSBML.

One another tool worth mentioning is Saint, which provides an automated SBML annotation environment. Further information can be found on its SourceForge project.

Annotation guidelines (for curators)

More detailed guidelines for model annotation, with specific examples, are available; although those have been primarily designed for our curators.