Protein–Protein Interface Topology as a Predictor of Secondary Structure and Molecular Function Using Convolutional Deep Learning

To power the specific recognition and binding of protein partners into functional complexes, a wealth information about structure function is necessarily encoded global shape protein–protein interfaces their local topological features. identify whether this case, study uses convolutional deep learning methods (typically leveraged for 2D image recognition) on 3D voxel representations colored by burial depth. A novel two-stage network fed with voxelizations each interface at two distinct resolutions achieves balance between performance computational cost. From interfaces, tries to predict presence secondary motifs molecular corresponding complex. Secondary certain classes are found be very well predicted, validating hypothesis that conveyor higher-level information. Interface patterns triggering also identified described.