Integration of transcription factor binding and gene expression by associative clustering
Janne Nikkilä, Christophe Roos and Samuel Kaski
In: KRBIO05, Symposium on Knowledge Representation in Bioinformatics, Espoo, Finland(2005).

Abstract

We integrate paired genomic data sets to reveal their dependencies. We suggest using a dependency-\-max\-i\-mizing clustering method for the task. The recently introduced method {\it associative clustering (AC)} finds groupings of genes for which the two data sources are maximally dependent. The dependencies between data sources become represented as a contingency table, which is optimized to reveal the association between data sets, bypassing the possible incommensurability between the data sets. The method is applied to searching for regulatory interactions in yeast, by looking for dependencies between gene expression profiles and regulator binding patterns.

[Edit]