Probabilistic Framework for Transcription Factor Binding Prediction

We formulate a probabilistic framework for transcription factor (TF) binding prediction that is built on the standard position specific frequency matrix (PSFM) and higher order Markovian background models. Contrary to the traditional hypothesis testing based methods which report a significance (p) value of TF binding at every possible base pair position in a promoter sequence, we develop a probabilistic methodology to assess TF binding to whole promoter sequences. Performance of the proposed method is demonstrated via simulations.