Information Provided by Ravi Rao
Additional Details
Computation of conservation laws must be carried out in order to obtain a non-singular Jacobian of a biochemical system. This allows the size of the model to be reduced so that only independent species have to be simulated. While current simulators are excellent at computing the conservation relations for small networks, the complexity of large networks and the numerical methods employed by them allow numerical errors to accumulate making it very difficult to generate reliable conservation laws for large networks which involve hundreds or even thousands of species and reactions.
I have successfully developed a new numerical method that uses Householder QR algorithm to identify the dependent and independent species in an efficient manner. This method is highly robust and correctly identifies all the conservation laws for large networks that other methods simply cannot handle. Furthermore, I have integrated this method into the SBW for other applications to use in simplifying their simulations. It is available as a module edu.kgi.StructAnalysis which can be downloaded from sys-bio.org (as part of SBW 2.5 and above). This algorithm has been implemented by a number of other tools, an example of which is Oscill8, the bifurcation analysis tool. Users can load a network into JDesigner and access Oscill8 to investigate possible bifurcations – Oscill8 calls the XPP translator which in turn calls the Conservation analysis module to identify the conservation laws, and subsequently converts the model to XPP format for use by Oscill8.
I have also developed a graphical user interface to the main computational engine that computes the conservation laws for large biochemical networks (given as SBML models). This application also allows for computation of many of the other relevant structural analysis matrices and also performs five tests on the generated conservation laws to ensure their accuracy. In addition to the QR algorithm, the tool also can perform the conservation analysis using LU partial pivoting or LU full pivoting of the stoichiometry matrix. Screen shots of graphical user interface for Structural Analysis tool (called Structural Analysis GUI), are shown below, in the normal and visual modes where the connectivity structure in the stoichiometry matrix can be inferred.
The core algorithm central to this work was published as an original article in the Feb 2006 issue of Bioinformatics Journal. Vol.22 (3), 346-353 (Title: Conservation analysis of large biochemical networks). This study was also a part of a poster (Title: Next generation computational tools for biochemical network models) presented at the DOE Joint Genomics: GTL Workshop, Feb 12-15, 2006, in Washington, D.C.,
A notable observation is that this algorithm has now been implemented by all major simulation tools for biochemical networks including COPASI, PySCeS, and recently by MathWorks (in Simbiology 2.0).
Collaborators: H.M.Sauro and V. Chickarmane (KGI), Emery Conrad(Virgina Tech University) and Brett Olivier (Stellenboch University, South Africa)
Please click here for additional details on Conservation analysis of biochemical networks.
 |
| Graphical Interface to Conservation Analysis |
 |
| Re-ordered stoichiometry matrix in visual mode |
