End-To-End Prediction Of Protein-Protein Interaction Based On Embedding And Recurrent Neural Networks

Protein-Protein interactions (PPIs) are key to many important life processes, such as cancer replication or DNA transcription. While in vivo or in vitro methods for PPI screening exist, they are expensive and computational approaches have been proposed to address PPI prediction. Previous computational methods rely on hand crafted features to capture the underlying information of the protein data. In this work we present a deep neural network architecture leveraging embedding techniques and recurrent neural networks to extract features and predict interaction between protein pairs. The results achieved are similar to those obtained by other state-of-the-art computational approaches to the problem but without any feature engineering involved, directly using the raw amino acid sequences.