Coding nucleic acid sequences with graph convolutional network

Genome sequencing technologies reveal a huge amount of genomic sequences. Neural network-based methods can be prime candidates for retrieving insights from these sequences because of their applicability to large and diverse datasets.However, the highly variable lengths of nucleic acid sequences severely impair the presentation of sequences as input to the neural network. Genetic variations further complicate tasks that involve sequence comparison or alignment. Here, we propose a graph representation of nucleic acid sequences called gapped pattern graphs. These graphs can be transformed through a Graph Convolutional Network to form lower-dimensional embeddings for downstream tasks. On the basis of the gapped pattern graphs, we implemented a neural network model and demonstrated its performance in studying phage sequences. We compared our model with equivalent models based on other forms of input in performing four tasks related to nucleic acid sequences--phage and ICE discrimination, phage integration site prediction, lifestyle prediction, and host prediction. Other state-of-the-art tools were also compared, where available. Our method consistently outperformed all the other methods in various metrics on all four tasks. In addition, our model was able to identify distinct gapped pattern signatures from the sequences.