Enhancer Prediction with Histone Modification Marks Using a Hybrid Neural Network Model

Abstract

An enhancer is a regulatory region in DNA, increasing transcription of a gene by combining with transcription factors. Whereas a promoter is located near a transcription start site, an enhancer can often be located far from a target gene, making hard to identify enhancer regions in DNA. Therefore, many researches in Bioinformatics have challenged to classify enhancer region computationally. In this paper, a hybrid neural network, Convolutional Neural Network followed by Recurrent Neural Network, are used for classifying enhancer regions in DNA with histone modification marks input and my model showed high performance in evaluation. With the trained model, optimizing virtual input matrix can give insight into how histone modification marks represent enhancer regions.