Computational Modeling of Interactions in Dynamic Biological Networks

By Mingzhou (Joe) Song, NMSU

Mar 24, 2009

CNLS Conference room.

Modeling dynamic interactions among genes and environment can lead to a quantitative understanding of mechanisms in cellular processes, such as transcription regulation, metabolism, and disease pathways. The computational problem of identifying network models to account for temporal dependencies among interacting genes and environmental stimuli from high-throughput gene expression data is addressed. The discrete dynamical system modeling and the generalized logical network modeling will be introduced and compared. In both frameworks, the detection of interactions is based on statistical significance of hypothesis testing on strength of associations. Such an approach allows one to control false-positive rates explicitly. Simple but representative simulation studies demonstrate the advantage of the statistical approaches to network topology recovery, over dynamic Bayesian networks. The modeling will be explained in two examples. The first reconstructs discrete dynamical system models for gene regulatory networks in ethanologenic yeast Saccharomyces cerevisiae in response to 5-hydroxymethylfurfural, a bioethanol conversion inhibitor. The second reconstructs generalized logical networks for gene regulatory network in mouse brain in response to alcohol treatment.

Back to CNLS q-bio Seminars.