Learning from Protein Structure with Geometric Vector Perceptrons (2020-09-03T00:00:00.000000Z)

TL;DR

This work introduces geometric vector perceptrons, which extend standard dense layers to operate on collections of Euclidean vectors and improves over existing classes of architectures, including state-of-the-art graph-based and voxel-based methods.

Abstract

Learning on 3D structures of large biomolecules is emerging as a distinct area in machine learning, but there has yet to emerge a unifying network architecture that simultaneously leverages the graph-structured and geometric aspects of the problem domain. To address this gap, we introduce geometric vector perceptrons, which extend standard dense layers to operate on collections of Euclidean vectors. Graph neural networks equipped with such layers are able to perform both geometric and relational reasoning on efficient and natural representations of macromolecular structure. We demonstrate our approach on two important problems in learning from protein structure: model quality assessment and computational protein design. Our approach improves over existing classes of architectures, including state-of-the-art graph-based and voxel-based methods.

Authors

Bowen Jing

3 papers

Stephan Eismann

3 papers

Patricia Suriana

3 papers

TL;DR

Abstract

Authors

References50 items

Hierarchical, rotation‐equivariant neural networks to select structural models of protein complexes

Fast and Flexible Protein Design Using Deep Graph Neural Networks.

GraphQA: protein model quality assessment using graph convolutional networks

A Review of Deep Learning Methods for Antibodies

ProteinGCN: Protein model quality assessment using Graph Convolutional Networks

Directional Message Passing for Molecular Graphs

DenseCPD: Improving the Accuracy of Neural-Network-Based Computational Protein Sequence Design with DenseNet

Protein sequence design with a learned potential

ProDCoNN: Protein design using a convolutional neural network

Fast and flexible design of novel proteins using graph neural networks

Deciphering interaction fingerprints from protein molecular surfaces using geometric deep learning

A structure-based deep learning framework for protein engineering

Geometric Graph Convolutional Neural Networks

Assessment of protein model structure accuracy estimation in CASP13: Challenges in the era of deep learning

QMEANDisCo—distance constraints applied on model quality estimation

Estimation of model accuracy in CASP13

Cormorant: Covariant Molecular Neural Networks

To Improve Protein Sequence Profile Prediction through Image Captioning on Pairwise Residue Distance Map

Deep convolutional neural networks for predicting the quality of single protein structural models

Generative Models for Graph-Based Protein Design

Smooth orientation-dependent scoring function for coarse-grained protein quality assessment

Enhanced Deep‐Learning Prediction of Molecular Properties via Augmentation of Bond Topology

Protein model quality assessment using 3D oriented convolutional neural networks

End-to-End Learning on 3D Protein Structure for Interface Prediction

Design of metalloproteins and novel protein folds using variational autoencoders

Relational inductive biases, deep learning, and graph networks

SPIN2: Predicting sequence profiles from protein structures using deep neural networks

Deep transfer learning in the assessment of the quality of protein models

Tensor Field Networks: Rotation- and Translation-Equivariant Neural Networks for 3D Point Clouds

Computational Protein Design with Deep Learning Neural Networks

Deep convolutional networks for quality assessment of protein folds

Attention is All you Need

VoroMQA: Assessment of protein structure quality using interatomic contact areas

Neural Message Passing for Quantum Chemistry

ProQ3D: improved model quality assessments using deep learning

Boosting Docking-Based Virtual Screening with Deep Learning

Inductive Bias of Deep Convolutional Networks through Pooling Geometry

Protein single-model quality assessment by feature-based probability density functions

Direct prediction of profiles of sequences compatible with a protein structure by neural networks with fragment‐based local and energy‐based nonlocal profiles

The conference paper

Et al

Macromolecular modeling with rosetta.

Residue centrality, functionally important residues, and active site shape: Analysis of enzyme and non‐enzyme families

Relating protein motion to catalysis.

Native protein sequences are close to optimal for their structures.

Approximation by superpositions of a sigmoidal function

Spatial Graph Convolutional Networks

and C

Biochemistry

Processing and evaluation of predictions in CASP4

Field of Study

Journal Information

Name

Volume

Venue Information

Name

Type

URL

Alternate Names