The influence of the background model on DNA motif prediction: An assessment for zinc finger transcription factor ZFX

Abstract

Motif finding is a computationally expensive procedure subject to noise and false positives, but of major importance in understanding gene expression and cancer. Several authors argued in favor of using higher order background models to better discriminate motifs. This paper studies the effect of using Markov higher order models in three commonly used algorithms to identify the ZFX transcription factor's binding sites from a mouse embryonic stem cells dataset. We conclude that there are particular Markov orders that yield improved outcomes for each algorithm.