MHGEnhancer:Predicting Enhancers Based on Multiple histone modification data

Enhancers play a critical role in gene regulation and are essential for maintaining cellular characteristics and functions. Accurate identification of enhancers aids in under-standing the regulatory mechanisms of gene expression and provides important insights for disease research. However, existing methods still face challenges in effectively and accurately identifying enhancers. To address this, our study proposes a novel enhancer identification method called MHGEnhancer. This meth-od is based on multiple histone modification data (H3K4me1, H3-K4me3, H3K27ac) and employs GraphSAGE as a base learner to capture the features of each histone modification, then integrate the prediction result of multiple histone modification models into a tensor and input it into the Full Connect Neural Network to generate the final prediction. Compared to traditional data integration methods, our approach demonstrates superior performance. Furthermore, MHGEnhancer significantly outperforms methods that rely on sequence features and single histone modification data in enhancer prediction. Overall, our research provides a new perspective for exploring enhancer identification.