BioModels.net Training Camp Day 1

Attached to the end of the BioSysBio conference, the 2nd BioModels.net Training camp is running from 13-15 January, 2007, with about 25-30 people in attendance The first day involved a series of talks, followed by a discussion of how to increase the amount of collaboration between BioModels.net and other organizations (among other things). After the work finished, there was a lively meal at a nearby italian restaurant (which included, partway through dinner, a travelling troupe of accordionist, guitarist and tambourinist – though the tambourine was mainly used for collecting donations!). What follows is a summary of what was discussed on Day 1.

Talks:

  • Nicolas Le Novère – BioModels projects. The EBI side
    • MIRIAM (Minimum Information Requested In the Annotation of biochemical Models) was launched in 2004, and has three parts:
      • reference correspondence. Your model must be in a public machine-readable format and must be described in a single location (this is the reference description). The structure of the model must reflect the biological processes listed in the reference description. All quantitative attributes must be defined, and finally the model should get the same result as that described in the reference description.
      • attribute annotation. Your model must be named, and the reference description must have a citation. Both the authors of the model and the model creator must be credited, and the date and time of creation and last modification, as well as a statement about the terms of distribution must be included.
      • external resource annotation. To unambiguously relate a piece of knowledge to a model constituent, the referenced info must be described using a triplet of data-type, identifier and qualifier. To aid in this process, the community must agree on a set of standard valid data types.
    • The MIRIAM database at the ebi stores a list of data resources and their URIs. It is accessible via web browser and via web services, and via xml download.
  • Michael Hucka – SBML, a summary of current happenings and plans
    • The ability to exchange models is critical, and requires a common file format (which didn’t happen until 2000).
    • SBML is defined and created using UML, and then released as XML schema. Therefore, it is targeted at XML but mostly independent of it. Intended to be machine-readable, not human-readable.
    • SBML Levels are meant to co-exist. Each level introduces new features and enriches existing features. In this way, it gains power at the expense of simplicity.
      • Level 1 is mostly basic compartmental modeling.
      • Level 2 has more features including user-defined functions, events, types, initial conditions and constraints.
      • Level 3 in development.
    • SBML now the de facto standard, and supported by >100 systems, accepted by journals including Nature, BMC and PloS. It is also used in textbooks (such as Darren Wilkinson’s book on Stochastic Modelling in Systems Biology) and courses.
    • It is important to capture meaning in a model by associating ontology terms to objects within the SBML model. When not using ontologies, the terms are completely unregulated and not very useful for programmatic analysis.
  • Herbert Sauro – Human-readable model definition language (and Frank Bergmann)
    • Meant to generate new sbml rather than combine existing SBML models.
    • A model is made up of modules, which define a rate process. A module describes a process by a set of one or more reactions. State variables inside a module are either local or accessible through the module interface.
    • Targeting synthetic biology people as well as systems biology people. For biologists and computer scientists and engineers.
    • (“Editor’s” note: some from these groups may not be immediately comfortable with this language, which looks a little like pseudo-code.)
  • Lena Strömbäck Standards, exchange and databases
    • Provided an overview of her work on comparison of all currently-used data exchange formats for molecular interactions, including BioPax, PSI MI and SBML.
  • Dagmar Koehn Working with different pathway standards
    • There are already applications out there which perform various conversions relevant to the systems biology modelling world, including:
      • comparing xml schemas: Clio, COMA++
      • comparing owl models: SAMBO, COMA++, Protege
      • comparing different formats: some converters for XML to OWL, PSI MI to bioPAX.
    • She has worked on creating an application to perform this third comparison.
    • Has discovered that transformation is only possible in one direction, from xml schema to owl model, due to the fact that OWL is a semantically-aware format, while xml is just syntactic. She is developing an application that can deal with XML to OWL to XML again.

The discussion covered the possibility of talking with the FuGE and OBI people on linking SBML models to the experiments they are connected to; the need for a formal way to describe the simulation process, as “run the simulation for 1000 sec” can give completely different results to “run the simulation for 1 sec”; the possibility of using evidence tagging for annotation sections of SBML by extending the GO annotation attribution as the people at Newcastle University are doing.

Advertisements

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s