Structural Learning of Proteins Using Graph Convolutional Neural Networks

The exponential growth of protein structure databases has motivated the development of efficient deep learning methods that perform structural analysis tasks at large scale, ranging from the classification of experimentally determined proteins to the quality assessment and ranking of computationally generated protein models in the context of protein structure prediction. Yet, the literature discussing these methods does not usually interpret what the models learned from the training or identify specific data attributes that contribute to the classification or regression task. While 3D and 2D CNNs have been widely used to deal with structural data, they have several limitations when applied to structural proteomics data. We pose that graph-based convolutional neural networks (GCNNs) are an efficient alternative while producing results that are interpretable. In this work, we demonstrate the applicability of GCNNs to protein structure classification problems. We define a novel spatial graph convolution network architecture which employs graph reduction methods to reduce the total number of trainable parameters and promote abstraction in interme-diate representations. We show that GCNNs are able to learn effectively from simplistic graph representations of protein structures while providing the ability to interpret what the network learns during the training and how it applies it to perform its task. GCNNs perform comparably to their 2D CNN counterparts in predictive performance and they are outperformed by them in training speeds. The graph-based data representation allows GCNNs to be a more efficient option over 3D CNNs when working with large-scale datasets as preprocessing costs and data storage requirements are negligible in comparison.