Varela Rey, M;
Varela Rey, M;
- view fewer
A new framework for the integration of models in biology.
Presented at: UNSPECIFIED.
The goal of Systems Biology is to understand large-scale biological systems. The problem often is tackled by building and composing models of biological processes at many different spatio-temporal scales, such as the behaviour of a complete organ derived from the molecular behaviour inside cells. This presents two challenges: integration of heterogeneous biological models and curation of models with the experimental data used to validate them. We aim to develop a robust and scalable approach for model construction and integration, which enables heterogeneous model integration, while allowing a wide variety of models to be executed using the most convenient software tool. The work is part of a DTI-funded Beacon Project to construct an in-silico liver (pizza.cs.ucl.ac.uk/grid/biobeacon), an organ of great medical importance with a relatively homogeneous structure and a dominant cell type, the hepatocyte. We model elements involved in biological model construction and validation to create a biological meta-model (Atkinson & Kuhne, 2003, Finkelstein et al, 2004). This relates the model, the modelling scheme, embedded assumptions and parameter values or experimental results used to 'run' or interpret the model. It includes results or interpretations obtained from the model, and the software environment. We use the meta-model to develop a new computational framework consisting of component middleware (Foster et al, 2002) and supporting services for integrating existing, heterogeneous, models. The middleware includes wrappers for software modelling tools, that supply a set of standard interfaces; smart connectors for building composite models; and a coordinator, or workflow execution service, that allows the models to be invoked appropriately. Connectors can, for example, solve numerically a composite model, in which some sub-models form a feedback loop. Thus far we have Mathematica and Xppaut wrappers for ODE based models. Our middleware supports the instantiation of model components using existing modelling tools, and enables communication between components, repositories of experimental data and existing interpretations. The framework supports the execution of composite models built from several sub-models. Parameters for models are obtained from a repository. Each parameter used in our models is documented according to a detailed schema, with particular attention to linking the parameter information to its experimental basis. The extensible markup language XML is used so that we are able to take advantage of existing tool support, allowing us to browse and present the information in an efficient and accessible manner. We can also perform queries on the data, allowing us to find parameters, for example, based on the work of a particular author, or referring to a specific biological entity. Eventually, this parameter database will form part of a context service that will allow models to automatically gain access to whatever parameter information is appropriate to a specific modelling application. The results, or interpretations of the models, are stored by the interpretation service. A model repository is used to store existing models, and can be systematically searched to find desired sub-models for creation of new composite models. We will demonstrate the application of our middleware and wrappers with the computational integration of two models, one written in Xppaut, the other in Mathematica. We shall integrate a model of G-protein coupled receptor signalling with a model of calcium signalling pathways in hepatocytes, using parameters taken from Kummer (2002). We will also show further novel strategies for building large integrated models, illustrated by a linked model of the signalling and metabolic pathways associated with glycogenolysis in hepatocytes. Some experimental data to amplify the models will be included.
|Type:||Conference item (UNSPECIFIED)|
|Title:||A new framework for the integration of models in biology|
|UCL classification:||UCL > School of Life and Medical Sciences
UCL > School of Life and Medical Sciences > Faculty of Life Sciences
UCL > School of BEAMS > Faculty of Engineering Science
UCL > School of BEAMS > Faculty of Engineering Science > Engineering Science Faculty Office
UCL > School of BEAMS > Faculty of Maths and Physical Sciences
UCL > School of BEAMS > Faculty of Maths and Physical Sciences > Mathematics
Archive Staff Only