The key features of gene-regulatory networks are the interconnections between specific transcription factors (TF) and
cis-regulatory elements of the DNA. These connections function as an interface between signaling pathways and the regulation of gene
expression. After decades of intensive research effort only a small fraction of these connections is known.
A major and labour-intensive part of this effort is the characterization of the DNA-binding specificity of TFs
whose DNA-binding domain enables the specific recognition of short DNA motives in the promoter region of their proximal target genes.
To further increase our knowledge about the specific interactions between TFs and cis-regulatory elements, we designed
an algorithm which allows for predicting the binding specificity of transcription factors with high accuracy. Apparently, the protein sequence and structure of
the DNA-binding domain determines its function, which is the molecular recognition and binding to a defined set of DNA motives.
Thus we approach to estimate the binding specificity of TFs based on structural, physicochemical and properties of their DNA-binding domains.
Employing support vector regression we estimate the similarity of the binding specificities of two TFs based on diverse features incorporating
domain sequence similarity, secondary structure and phylogenetic distance. This approach provides a quantitative measure for the functional
similarity of two factors and enables the transfer of DNA-binding specificity data with low error.
|
