Center for Bioinformatics at KU

As part of our public Seminar Series
http://bioinformatics.ku.edu/seminars

May 23, Mon, 2005
4:00pm - 5:00pm, rm 1005, Haworth

Hui Lu

Structural Bioinformatics Prediction of Protein-DNA interactions

Transcription factors (TF) recognize specific DNA promoter sequences and initiate mRNA transcription to regulate spatial and temporal patterns of gene expression. Identifying transcription factors and their target sites is essential to understanding gene regulatory networks. Here, we present a systematic protocol comprised of three components aimed at predicting transcription factors, DNA binding sites, and gene regulation using structural bioinformatics approaches. First, we identify the transcription factor through machine learning techniques using various features such as electrostatics, amino acid composition, and structural information. Second, we build the complex structure by applying machine learning based binding site prediction and docking. Third, we developed a statistical potential describing the protein DNA interaction propensity by including distance dependency, a multi-body term, and a spatial distribution. These potentials are used to examine the compatibility between the TF and the corresponding DNA binding site in the promoter region. In each of the above steps, we provide and apply a benchmark to assess the accuracy of prediction. Overall, our protocol provides a tool to predict the TF, the genes they are regulating, and the specific binding sites.