Fast end-to-end learning on protein surfaces

By Freyr Sverrisson, Jean Feydy, Bruno E. Correia, Michael M. Bronstein. 2020

197 citations recorded by the source corpus.

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.

Subfields: Computational Mechanics, Industrial and Manufacturing Engineering, Materials Chemistry

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Mechanics

In Industrial and Manufacturing Engineering

In Materials Chemistry

Citations in the corpus

EquiBind: Geometric Deep Learning for Drug Binding Structure Prediction (2022, background): “Gainza et al. (2020); Sverrisson et al. (2021) leverage geometric deep learning and mesh convolutional neural networks (CNN) to embed protein surface patches into ﬁngerprints and allow for fast scanning and binding site identiﬁcation, removing the need for handcrafted or expensive pre-computed…”
Protein Representation Learning by Geometric Structure Pretraining (2022, methodology): “Another potential direction is to extract features from protein surfaces (Gainza et al., 2020; Sverrisson et al., 2021; Dai & Bailey-Kellogg, 2021).”
P HYSICS - INFORMED DEEP NEURAL NETWORK FOR RIGID - BODY PROTEIN DOCKING (2022, methodology): “In this paper, we have presented an extension to our previous method dMaSIF (Sverrisson et al., 2021), which allowed us to perform rigid-body docking of protein-protein pairs.”
ScanNet: an interpretable geometric deep learning model for structure-based protein binding site prediction (2021, background): “as sequences [22, 23], residue graphs [24–27], atomic density maps [28–34], atomic point clouds [35] or molecular surfaces [36, 37], each capturing different functionally relevant features.”
Artificial intelligence in drug development (2025, unknown): “…interest in deep learning models 42,55 , and these models may outperform traditional scoring functions by handling both three-dimensional structural 56 and nonstructural data 57–64 , their performance heavily depends on ligand pose accuracy and is primarily suitable for known receptor structures.”
GeoMol: Torsional Geometric Generation of Molecular 3D Conformer Ensembles (2021, background): “…it is crucial to understand how a molecule binds to a speciﬁc target protein; this process heavily depends on the 3D structures of the two components, both in terms of geometric (shape matching) and chemical (hydrophobic/hydrophilic) interactions [Gainza et al., 2020, Sverrisson et al., 2020].”
Independent SE(3)-Equivariant Models for End-to-End Rigid Protein Docking (2021, methodology): “Following previous work (Sverrisson et al., 2021; Venkatraman et al., 2009), we deﬁne the surface of a protein point cloud ∈ { ∈ } − −|| − || The parameters σ and γ are chosen such that there exist no "holes" inside a protein (we found γ = 10 , σ = 25 to work well, see Appendix E).”