Identification of Transcriptional Promoter Sequence based on statistical filter bank model

This paper describes a new approach for locating transcription related signals, such as promoter sequence in nucleic acid sequences. Transcription Factor (TF) and corresponding polymerase binding to their DNA target site is a fundamental regulatory interaction. The most common model used to represent TF and polymerase binding specificities is a position weight matrix (PWM) [1], which assumes independence between binding positions. However, in many cases, this simplifying assumption does not hold. In this paper, we present a statistical filter model based on Chi-Square (χ2) distance [2], which is a statistical distance metric between the profiles of component vectors. It is a novel statistical method for modeling TF-DNA and polymerase-DNA interactions. Our approach also uses a generalized correlation algorithm to evaluate the combination coefficients for the filter bank. Simulation results show that the proposed approach identifies promoter sequences better than the PWM model method and Chi-Square (χ2) distance model.