DNA-GCN: Graph Convolutional Networks for Predicting DNA-Protein Binding

Predicting DNA-protein binding is an important and classic problem in bioinformatics. Convolutional neural networks have outperformed conventional methods modeling the sequence specificity of binding. However, none studies has utilized graph convolutional for motif inference. In this work, we propose to use We build a k-mer whole dataset based on co-occurrence relationship then learn DNA Graph Network(DNA-GCN) dataset. Our DNA-GCN initialized with one-hot representation all nodes, it jointly learns embeddings both k-mers sequences, as supervised by known labels sequences. evaluate our model 50 datasets from ENCODE. shows its competitive performance compared baseline model. Besides, analyze design several different architectures help fit datasets.