Informatik, TU Wien

Designing and optimising reaction fluxes and gene expression data in biological models

Metabolic engineering is a promising biotechnology approach with increasing demand of mathematical models for accurate design purposes. The goal is to overproduce metabolites of interest through genetic or metabolic intervention, as well as to identify non-native synthetic pathways.

Abstract

Metabolic engineering is a promising biotechnology approach with increasing demand of mathematical models for accurate design purposes. The goal is to overproduce metabolites of interest through genetic or metabolic intervention, as well as to identify non-native synthetic pathways.

We propose a methodology suitable for optimal design and assessment of biological models. More specifically, we adopt single- or multi-objective optimisation techniques driven by epsilon-dominance and identifiability analysis. Our novel optimisation algorithm searches for the values of the gene expression profiles and the flux rates that optimise multiple cellular functions simultaneously. As a result, we obtain a Pareto front consisting of a set of strains specialised in the concurrent production of all the objectives chosen. Furthermore, we are able to quickly implement the effect that the underexpression or overexpression of specific genes has on the Pareto front, an approach widely used to link metabolism and malignancy in tumour biology. The systematic analysis of the Pareto surface emerging from the biological system can be reached by evaluating the sensitivity of its components, while the functional relations can be inferred from its constraints. To this end, we adopt sensitivity and identifiability analyses. Finally, the robustness analysis proves useful to assess how robust is a Pareto optimal solution after a perturbation in the model.

Our framework can be used also to optimise reaction fluxes and gene profiles simultaneously. The condition of Pareto optimality can be relaxed (e.g., with epsilon-dominance) to include suboptimal points that boost the convergence of the algorithm. Our approach makes it possible to design, investigate, optimise, and cross-compare biological models built with various techniques, including ordinary differential equations, differential algebraic equations, flux balance analysis and gene-protein reaction mappings.

Biography

Claudio Angione is a PhD student in the Artificial Intelligence group of the Computer Laboratory, University of Cambridge. He received the BSc degree in Mathematics in July 2009 from the University of Catania, with the thesis "Chaotic Maps and Bose—Einstein Condensation for Algorithms and NP-Complete Problems". In July 2011 he received the MSc degree in Mathematics from the University of Catania, with the thesis "Phase Transition Detection and Model Checking using Bose—Einstein Condensation in Random Satisfiability Problems". In January 2012 he was awarded the "Anile Prize 2011" for the best thesis at the Department of Mathematics and Computer Science of the University of Catania. In April 2013 he received the MSc diploma from the Institute for Advanced Studies of the University of Catania, with the thesis "Improving Satisfiability-solvers using Classical and Quantum Statistical Distributions".

His current research interests include systems biology, multi-objective optimisation, organelle models, computation with molecular machines, and statistical-physical approaches in satisfiability problems.

Note

This talk is organized by the Cyber-Physical Systems Group at the Institute of Computer Engineering.