Abstract: |
A novel computational method (called p53HMM) is presented that utilizes Profile Hidden Markov Models (PHMM's) to estimate the relative binding affinities of putative p53 response elements (RE's), both p53 single-sites and cluster-sites. These models incorporate a novel ``Correlated Baum Welch'' training algorithm that provides increased predictive power by exploiting the redundancy of information found in the repeated, palindromic p53-binding motif. The predictive accuracy of these new models are compared against other predictive models, including position specic score matrices (PSSM's, or weight matrices). Finally, we provide experimental evidence that verifies a predicted p53-target site that regu- lates the CHMP4C gene. The P53HMM algorithm is available on-line from http://tools.csb.ias.edu. |