Exploring the independence of gene regulatory modules
Janne Nikkilä, Antti Honkela and Samuel Kaski
In: Probabilistic Modeling and Machine Learning in Structural and Systems Biology (PMSB 2006), 17-18 June 2006, Tuusula, Finland.
We study the discovery of gene regulatory modules based on
transcription factor (TF) binding data and expression data from gene
knockouts. We invoke the natural assumption that regulatory modules
predominantly operate independently, which makes it possible to
apply a new method for extracting them: the Independent Variable
Group Analysis. We demonstrate that i) the independence assumption
helps in discovering the regulatory modules from TF data, and ii)
the independent gene modules discovered from TF-data can be found
also in expression data from gene knockouts. This demonstrates that
the regulatory effects by transcription factors are observable in
knockout experiments. It additionally suggests that the difficult
interpretation of the knockout experiments could be eased by taking
into account the independent regulatory modules.