End-to-End Learning on 3D Protein Structure for Interface Prediction (2018-07-03T00:00:00.000000Z)

TL;DR

The first end-to-end learning model for protein interface prediction, the Siamese Atomic Surfacelet Network (SASNet), is developed and it is found that SASNet outperforms state-of-the-art methods trained on gold-standard structural data, even when trained on only 3% of the new dataset.

Abstract

Despite an explosion in the number of experimentally determined, atomically detailed structures of biomolecules, many critical tasks in structural biology remain data-limited. Whether performance in such tasks can be improved by using large repositories of tangentially related structural data remains an open question. To address this question, we focused on a central problem in biology: predicting how proteins interact with one another—that is, which surfaces of one protein bind to those of another protein. We built a training dataset, the Database of Interacting Protein Structures (DIPS), that contains biases but is two orders of magnitude larger than those used previously. We found that these biases significantly degrade the performance of existing methods on gold-standard data. Hypothesizing that assumptions baked into the hand-crafted features on which these methods depend were the source of the problem, we developed the first end-to-end learning model for protein interface prediction, the Siamese Atomic Surfacelet Network (SASNet). Using only spatial coordinates and identities of atoms, SASNet outperforms state-of-the-art methods trained on gold-standard structural data, even when trained on only 3% of our new dataset. Code and data available at https://github.com/drorlab/DIPS.

Authors

Patricia Suriana

3 papers

Raphael J. L. Townshend

4 papers

R. Dror

5 papers

TL;DR

Abstract

Authors

References38 items

Variational Coarse-Graining for Molecular Dynamics

Capsule Networks for Protein Structure Classification and Prediction

BIPSPI: a method for the prediction of partner-specific protein–protein interfaces

3D Steerable CNNs: Learning Rotationally Equivariant Features in Volumetric Data

Graph Convolutional Policy Network for Goal-Directed Molecular Graph Generation

N-body Networks: a Covariant Hierarchical Neural Network Architecture for Learning Atomic Potentials

3D Molecular Representations Based on the Wave Transform for Convolutional Neural Networks.

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

Deep convolutional networks for quality assessment of protein folds

Protein Interface Prediction using Graph Convolutional Networks

DeepSite: protein‐binding site predictor using 3D‐convolutional neural networks

IntPred: a structure-based predictor of protein–protein interaction sites

SchNet: A continuous-filter convolutional neural network for modeling quantum interactions

3D deep convolutional neural networks for amino acid environment similarity analysis

Neural Message Passing for Quantum Chemistry

TopologyNet: Topology based deep convolutional and multi-task neural networks for biomolecular property predictions

Atomic Convolutional Networks for Predicting Protein-Ligand Binding Affinity

ANI-1: an extensible neural network potential with DFT accuracy at force field computational cost

Pushing the Backbone in Protein-Protein Docking.

Quantum-chemical insights from deep tensor neural networks

Molecular graph convolutions: moving beyond fingerprints

A hybrid method for protein–protein interface prediction

AtomNet: A Deep Convolutional Neural Network for Bioactivity Prediction in Structure-based Drug Discovery

Convolutional Networks on Graphs for Learning Molecular Fingerprints

Updates to the Integrated Protein-Protein Interaction Benchmarks: Docking Benchmark Version 5 and Affinity Benchmark Version 2.

Simulated unbound structures for benchmarking of protein docking in the Dockground resource

Progress and challenges in predicting protein interfaces

Protein-ligand binding site recognition using complementary binding-specific substructure comparison and sequence profile alignment

3did: a catalog of domain-based interactions of known three-dimensional structure

Protein data bank

Partner-Aware Prediction of Interacting Residues in Protein-Protein Complexes from Sequence Data

Generalized neural-network representation of high-dimensional potential-energy surfaces.

Prediction‐based fingerprints of protein–protein interactions

Flexible protein-protein docking.

Twilight zone of protein sequence alignments.

Signature Verification Using A "Siamese" Time Delay Neural Network

Protein-Ligand Scoring with Convolutional Neural Networks

Supplementary information Predicting protein-protein interface residues using local surface structural similarity

Field of Study

Venue Information

Name

Type

URL

Alternate Names