NOTICE: You are viewing a page of the openwetware wiki. Our "dewikify" feature makes a wiki page appear as a normal web page. On September 22nd 2017, this feature will GO AWAY and this URL will redirect to the source URL on our wiki. We're sorry for the inconvenience.

Chemical and Biological Systems Engineering Laboratory

Home                    Research                    People                    Publications                    Internal                    News       

Sridharan Srinath

Sridharan Srinath

Department of Chemical and Biomolecular Engineering
4 Engineering Drive 4 Block E5 #B-05
National University of Singapore
Singapore 117576
Tel:+65 6516 7859

I work in the Gunawan lab at National University of Singapore.


Research Interests

Model Identification in the Biochemical Systems Theory

Recent advances in technology permit high throughput experiments at genomic, transcriptomic, proteomic and metabolomic levels. The information obtained from time-series data, however, is implicit and requires extensive data analysis. Mathematical modeling of biological systems has found increasing applications for investigating dynamics in complex cellular processes and has given rise to a new field called systems biology. In systems biology, biological processes like signal transduction and metabolic networks are often modeled using differential equations. These models often include many unknown parameters like enzymatic reaction rate constants, which are to be determined by fitting to time-course experimental data. Using the power-law formalism, the Biochemical Systems Theory (BST) coupled with high-throughput biological measurements transform the model identification into an inverse problem of estimating model parameters from experimental data.

Given time-series data and a model, parameter estimation can be thought as the “inverse problem” of generating predictions from model. Despite the large number of publications on this topic, this task remains the bottleneck in the application of BST modeling in biologically related area. Many studies in the literature have focused on developing comprehensive parameter estimation techniques that exploit many of the mathematical features of canonical models within the BST, such as S-systems or generalized mass action (GMA) or lin-log models. However, many challenges arise from the same underlying problem; incomplete and noisy measurements lack the necessary information in order to accurately estimate the model parameters. This is a parameter identifiability problem. Thus, the focus of this work is to investigate the identifiability of metabolic network models, and to suggest model refinement or experimental design that maximizes the number of estimable parameters from data.

Two types of identifiability property are considered. First, a priori identifiability analysis yields the identifiable parameters under the assumption of noise-free data. Parametric sensitivities are used as a basis for selecting the a priori identifiable parameters. Secondly, practical identifiability gives the identifiable parameters when the data are contaminated with noise. In other words, this analysis gives the accuracy with which the parameters can be estimated. The practical identifiability analysis methods are based on linear(ized) and nonlinear regression analysis, particularly the statistical inference of confidence interval or region of the parameter estimates. The applications to two inverse modeling problems within the BST point to the lack of parametric identifiability as the root cause of the difficulty faced in the inverse modeling. Although this work focuses on the BST models, the analyses can be applied to other types of models, and the issue of parameter identifiability is expected to be a common problem in other biological modeling.

Key words: identifiability analysis, inverse modeling, Biochemical Systems Theory, confidence region

Constrained-based modelliing
Metabolic Engineering

Journal Articles

Peer reviewed Conference Proceedings

Oral & Poster Presentations

Useful links