Fast end-to-end learning on protein surfaces

Abstract

Proteins’ biological functions are defined by the geometric and chemical structure of their 3D molecular surfaces. Recent works have shown that geometric deep learning can be used on mesh-based representations of proteins to identify potential functional sites, such as binding targets for potential drugs. Unfortunately though, the use of meshes as the underlying representation for protein structure has multiple drawbacks including the need to pre-compute the input features and mesh connectivities. This becomes a bottleneck for many important tasks in protein science.In this paper, we present a new framework for deep learning on protein structures that addresses these limitations. Among the key advantages of our method are the computation and sampling of the molecular surface on-the-fly from the underlying atomic point cloud and a novel efficient geometric convolutional layer. As a result, we are able to process large collections of proteins in an end-to-end fashion, taking as the sole input the raw 3D coordinates and chemical types of their atoms, eliminating the need for any hand-crafted pre-computed features.To showcase the performance of our approach, we test it on two tasks in the field of protein structural bioinformatics: the identification of interaction sites and the prediction of protein-protein interactions. On both tasks, we achieve state-of-the-art performance with much faster run times and fewer parameters than previous models. These results will considerably ease the deployment of deep learning methods in protein science and open the door for end-to-end differentiable approaches in protein modeling tasks such as function prediction and design.

Unconstrained generation of synthetic antibody-antigen structures to guide machine learning methodology for real-world antibody specificity prediction

“…Specifically, beyond detailed benchmarking of already implemented encodings, investigations of graph-based encoding, physicochemical properties, full torsion angle, or 3D voxels, the addition of physics-based priors may be needed (114). Furthermore, comparative insights into different ML architectures such as Transformers, graph-based ML (35,115,116) as well as attribution strategies such as integrated gradients to map the models' prediction to the underlying biology are warranted (114). These follow-up investigations are especially important for addressing the transportability of trained ML architectures to other epitopes or antigens (117).…”

“…Machine learning (ML) represents an increasingly used approach for antibody-antigen binding prediction (3,10,18,(21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32) given its capacity to infer the hidden nonlinear rules underlying high-complexity protein-protein interaction (30,(33)(34)(35)(36)(37)(38). Recent reports suggest that binding prediction based on antibody sequence or structure may be feasible as long as sufficiently large antigen(epitope)-specific antibody datasets become available (15,16,(39)(40)(41)(42)(43).…”

Deep graph learning of inter-protein contacts

“…layers of 128 hidden units with BatchNorm and ReLU. Both coordinates and normals are used to represent the geometric properties of a monomer structure 23 . We standardize the geometric features so that they are invariant to the coordinate system used by the monomer structure.…”

“…However, it is slow in calculating docking maps and thus, cannot scale well to proteome-scale prediction. Some deep learning methods also use learned representations of tertiary structures, including voxels 20,21 and radial/point cloud representations on protein surfaces [22][23][24] . Meanwhile, some representations include anisotropy information in the structures 25,26 while others do not.…”

RosettaSurf - a surface-centric computational design approach

“…Having these features of proteins encoded in their surfaces facilitates powerful computational analysis. Recently, the machine learning-based approach MaSIF (molecular surface interaction fingerprinting) showcased the extraction of surface features from molecular surfaces to study protein structure-function relationships (14,15). Intuitively, the molecular surface forms the boundary of the protein and its surroundings, thus acting as the interface that engages in interactions with other molecules.…”