Mathematical models for biological regulation networks

Research themes:

The research activities of the research area will include:

1. Formulation of models of gene regulation network dynamics. For example, a model of the dynamics of protein concentration or transcription factors, codified by genes in a network, will be represented with equations which describe the interactions that regulate the synthesis and the degradation of some proteins from others. From a mathematical point of view, the models are described by differential non-linear ordinary equations and the regulation of molecular interactions are represented by functions depending from “edge” values which its numeric value is unknown and difficult to identify. The interactions can be (i) at a genic level, “switch-like”, or (ii) both at a genic and a metabolic level, with different temporal scales.

2. Development and implementation of methods for the analysis and quality simulation of the models of regulation networks defined at points (i) and (ii).

3. Development of quality methods for model assessment.

4. Application of the proposed methodologies for the study of biological regulation systems.

Objectives:

The aim of the research activity regards the definition and the development of a mathematical and computational environment for the mathematics modelling, analysis and quality simulation of biological regulation network models in which the interactions between different components of the network are regulated by “edge” values and are described by continuous non-linear functions of “sigmoidal” type.

This aim will be achieved through more specific sub-objectives which consist in the development of: quality simulation algorithms; codes for symbolic calculating; quality assessment model methods; “hybrid” methods of automatic construction of gene regulation networks capable of integrating representation methods of knowledge and genic expression learning methods from temporal series; methods to reduce the computational complexity of the simulation algorithms in case of networks with an elevated number of components, for example through the decomposition of the network in sub-networks, that are computationally tractable and independent from a logical and functional point of view. Application and evaluation of the proposed methodologies.

Return to the Machine Learning Research Area

Other Research Units in the Machine Learning Research Area:

Machine Learning for biological data

Bio-inspired learning