Context dependent models for discovery of transcription factor binding sites

Transcription factors play a crucial role in gene regulation, and the identification of transcription factor binding sites helps gain insight into gene regulatory mechanisms. The overall goal of this work is to describe a new method of binding site detection called Motif Discovery via Context Dependent Models (MDCDM). We characterize the motif (i.e., binding sites) by a series of position-dependent first-order Markov models. This model considers both the position-specific features of the motif and the dependence between positions of the motif. In addition, a “step-up” testing procedure is used to automatically determine the best-fitting Markov model for the background (i.e., nonsite regions). We compare our approach with the existing methods using both real and simulated data sets. The results show that the detection of binding sites can be greatly improved by accounting for dependence across positions in a motif and appropriately modeling the background dependence.