Bayesian Inference of Gene Regulatory Networks: From Parameter Estimation to Experimental Design

Datum

23.02.2012

Zeit

14:30 - 15:30

Sprecher

Johanna Mazur

Zugehörigkeit

Viroquant Research Group Modeling, Ruprecht-Karls-Universität Heidelberg

Serie

IMB - Seminar

Sprache

en

Hauptthema

Mathematik

Andere Themen

Biologie, Mathematik, Informatik

Beschreibung

The parameter estimation of differential equations models in systems biology is a demanding task due to a limited amount of data available, which contains additionally a lot of noise. Thus, technical problems like non-identifiabilities and sloppiness of parameters arise in inference tasks for differential equations models in systems biology. To address these difficulties properly, a Bayesian approach is adequate, since the noise in the data is captured properly, prior knowledge can be incorporated easily, and it offers distributions over parameters providing different parameter sets that are able to describe the data in a similar way. In this work, two new methods are presented, one for Bayesian parameter estimation and one for Bayesian experimental design. These methods were applied to a non-linear ordinary differential equations model describing the dynamics of gene regulation. Results which outperform existing methods will be shown for the DREAM2 Challenge #3 data.

Letztmalig verändert: 13.02.2012, 13:37:12