A new method for alleviating sequence-specific biases in DNase-seq

DNasel footprinting is an established approach for recognizing transcription factor binding sites. High-throughput DNase-seq arrays have been used to depict in vivo DNase footprints with single base pair resolution. Many different computational methods have been developed to predict binding sites through the DNase-seq footprinting. However, cleavage bias of DNaseI leads to a negative impact on the prediction accuracy. In this study, we present a new method to reduce the impact of this sequence-specific bias. In the experimental verification section, we use SVM to test our approach, the better classification result illustrates the effectiveness of our new method.