The Centre for Systems Biology at Edinburgh (CSBE) has developed a novel optimisation framework for systems biology models. At the meeting of the CSBE International Scientific Advisory Board in December 2009 the investigators of the centre demonstrated the framework to the advisory board members who responded very positively to the work. In particular one of the members of the advisory board, Prof Kitano, was sufficiently impressed by the SBSI optimisation framework that he suggested that it might be possible to integrate the optimisation framework with his own software research tool, CellDesigner. This has provided an exciting opportunity for the CSBE developers. It is possible that collaboration could begin between the centre in Edinburgh and the Systems Biology Institute (SBI) in Japan. This would lead to the development of an integrated analysis tool called CellAnalyser which incorporates the CellDesigner software and the optimisation framework developed at Edinburgh. Additional funding would be sought for this project, helping to sustain the research momentum developed by the Centre for Systems Biology at Edinburgh.
Who am I?:
Stephen Gilmore and Richard Adams, School of Informatics and Centre for Systems Biology at Edinburgh
The main outcome of the project was the creation of a version of the SBSI application, developed at the Centre for Systems Biology at Edinburgh (CSBE), which is compatible with CellDesigner, a graphical modelling tool for biological networks developed at the Systems Biology Institute (SBI) in Tokyo. This project confirmed that the CellDesigner platform’s plugin system is usable for the creation of a plugin that will encapsulate the functionality of SBSI and make it available to the CellDesigner user base.
CellDesigner is a widely used editor for drawing biological networks. It has the ability to save the network drawn in SBML format, the current de facto standard for biological models. SBSI is a framework for global parameter optimization of SBML models. SBSI uses parallel genetic algorithms and gives the user the ability to configure the optimization parameters. SBSI users can run the optimisation either locally on their own machine; or on a server; or on a massively parallel machine such as the UK’s high-performance HECTOR terrascale computing resource.
With the introduction of an SBSI plugin for CellDesigner, the user has the ability to use the SBSI parameter optimisation functionality either locally (on their own machine) or remotely (using the CSBE server). This allows users to optimise their CellDesigner model with minimal customization from the user’s part.
Do the outcomes match the objectives:
The outcomes match the original objectives expressed in the proposal. The project has achieved what it intended to achieve.
Benefits for the future:
The main benefit for the future is that modellers developing dynamic models of biological processes in CellDesigner will be able to easily and effectively parameterise their models using data from biological experiments. This will ensure that models are more accurate, and that predictions made using them are more dependable. The CellDesigner application has a substantial user base and the results of this project now allow these users to access the functionality of SBSI, to improve the quality of the models they create.
Opportunities for further research:
Having confirmed that the plugin mechanism is usable from our part, one of the most interesting opportunities for taking the research forward is the development of a more advanced plugin. This would be able to utilize more aspects of the CellDesigner platform and to deliver a product that includes more features of the SBSIVisual application. On top of that, with the development of the large-scale collaborative research project Garuda now beginning (http://www.systems-biology.org/~myukiko/garuda), SBSI has the chance to be an integral part of this project and therefore establish a leading role as the parameter optimization framework of choice for CellDesigner modellers.
Has the project created any new shared resources?:
The outcome of the project is a freely-available example plugin for CellDesigner which can be used by any users of the CellDesigner application. This version of the SBSI platform is deployed as a JAR file that once put in under the 'plugin' folder in the CellDesigner installation folder adds the designed functionality in the main CellDesigner program. Installation is simple and convenient even for non-expert users.
Future research plans:
The project led to the Edinburgh SBSI team being selected to host the third Garuda workshop. This will take place in the Informatics Forum in October 2010. Plans for follow-on projects and research funding were discussed at the second Garuda workshop and a session at the third workshop in Edinburgh will be devoted to future research plans including funding applications.
Publications and presentations:
The SBSI plugin for CellDesigner that was developed in this project was presented at the second Garuda workshop in Manchester on June 2010. The workshop was held in the National Centre for Text Mining,
Manchester Interdisciplinary Biocentre
, University of Manchester and was attended by approximately 30 members of the international systems biology community. The presentation elicited considerable interest from the delegates at the workshop and stimulated discussion of future directions for the work.
This project (and the resulting software) will enable us to utilize CellDesigner as a delivery platform for optimizing SBML models. One result of this will be exposing our SBSI parameter optimization framework a large user community that otherwise would be very difficult (if not impossible) to engage.
Use of funding:
There was no change of funding relative to the proposal. The funding awarded was spent as planned in the proposal. The project completed with a small surplus remaining in the budget.
How is it novel? What is exciting about it?:
The CellDesigner application is a major research effort in systems biology. It has an estimated 20,000 users, taking it considerably beyond the realm of a research prototype. Linking the SBSI optimisation framework with CellDesigner would expose the work of the Centre to a massive audience which it is unlikely to reach otherwise. Such opportunities come along rarely and we would like to make the most of this opportunity now that it has presented itself. Linking to the CellDesigner application would greatly promote the work of the Centre worldwide and bring significantly increased attention to the research programme which is underway.
What will I do next? What opportunities will it open up?:
Our next step will be to arrange a joint meeting with Prof Kitano’s group at their laboratory in Japan. This meeting will be held in February 2010 and will identify the most productive way to couple the two applications in order to lay the foundations for the forthcoming CellAnalyser tool. With respect to other research opportunities which this will open up, connecting the optimisation framework to the CellDesigner application enables the possibility of several other such integration efforts plugging software tools developed at Edinburgh into the widely-used CellDesigner application developed in Prof Kitano's Systems Biology Institute in Japan. This could lead to an ongoing programme of collaborative work funded via Anglo-Japanese funding routes.
What constitutes success? How risky is it?:
A successful outcome of this endeavour will be the provision of the SBSI optimisation framework as a plug-in for the CellDesigner application and widespread uptake of this plug-in within the CellDesigner community. This will require an initial period of study of the CellDesigner framework and its associated language, the Systems Biology Graphical Notation (SBGN). It will require original software development bridging between the SBGN language and the input of the SBSI optimisation framework. Undertaking software development with such a large user base available is a very risky activity because the large numbers of users of the CellDesigner application have developed a vast collection of biological models which will stress the capabilities of the SBSI optimisation framework. If the development work is done carefully and well then there is an opportunity to greatly increase the use of our tools. On the other hand if the development work is not properly funded and resourced then there is the danger of reducing the standing in which the Centre for Systems Biology at Edinburgh is currently held.
What resources do I bring to the project?:
The SBSI group at Edinburgh has a team of experienced developers who have crafted a very useful modelling and analysis environment. They have expertise in formal languages for systems biology ranging from SBML to Bio-PEPA. They have an existing application coded partly in Java and partly in C++ which will form the basis of the Edinburgh contribution to the CellAnalyser project. Although they do not yet have existing collaborative projects, members of CSBE have an ongoing relationship with Professor Kitano, which reduces the risk associated with the personal aspects of collaboration.
What resources and expertise do I need?:
The project will require six months of development effort from one of the SBSI developers. This can approximately be broken down into one month of initial study followed by three months of development effort followed by one month of testing and another final month of documentation. It is essential that the project is thoroughly tested and well documented in order that it can form a valuable contribution to the CellDesigner application.
What shared resources, if any, will the project create?:
The project will create original open source software components which can be used in a variety of other projects in CSBE and elsewhere. These will include parsers for the Systems Biology Graphical Notation and translators from this language into the input language of the SBSI optimisation framework. Potential users of these tools would include any member of CSBE or Informatics or the University's Division of Pathway Medicine or their collaborators. Edinburgh's contribution to the CellAnalyser project will be made available as open source software from the SBSI SourceForge project page. This decision to release software tools as open-source marks a change in practice from Professor Kitano’s lab because previous software releases have been closed-source. Thus IDEAlab funding will help to support the production of a valuable scientific research tool, made freely available to the research community.
What is the timescale?:
The project will take place over a timescale of six months. We anticipate that there will be a single developer working on the project throughout the six months, possibly with assistance from other members of the SBSI development team.