Accurate prediction of protein structures and interactions using a three-track neural network
Точное предсказание структур и взаимодействий белков с использованием трековой нейронной сети из трех каналов
2021-07-15
SCID: 54.1/mc55sf5t
Discuss with AI
1D-2D-3D information integrationprotein structure predictionprotein-protein complex modelingthree-track neural networkx-ray crystallography and cryo-electron microscopy modeling
Figures from the paper
Abstract (AI)
DeepMind presented notably accurate predictions at the recent 14th Critical Assessment of Structure Prediction (CASP14) conference. We explored network architectures that incorporate related ideas and obtained the best performance with a three-track network in which information at the one-dimensional (1D) sequence level, the 2D distance map level, and the 3D coordinate level is successively transformed and integrated. The three-track network produces structure predictions with accuracies approaching those of DeepMind in CASP14, enables the rapid solution of challenging x-ray crystallography and cryo-electron microscopy structure modeling problems, and provides insights into the functions of proteins of currently unknown structure. The network also enables rapid generation of accurate protein-protein complex models from sequence information alone, short-circuiting traditional approaches that require modeling of individual subunits followed by docking. We make the method available to the scientific community to speed biological research.
Key Findings
1
A three-track neural network integrating 1D sequence, 2D distance map, and 3D coordinate information achieves top performance among explored architectures.
2
The authors are releasing the method to the scientific community to accelerate biological research.
3
The method enables rapid solution of challenging x-ray crystallography and cryo-electron microscopy structure modeling problems.
4
The model rapidly generates accurate protein-protein complex models from sequence alone, bypassing separate subunit modeling and docking.
5
The network provides functional insights for proteins of currently unknown structure.
6
The three-track network produces structure predictions with accuracies approaching DeepMind's CASP14 results.
Research Object
Protein structures and protein–protein complexes predicted from amino-acid sequences
Research Subject
Accurate prediction of 3D protein structures and protein–protein interactions using a three-track neural network that integrates 1D sequence, 2D distance maps, and 3D coordinates
Publication Details
Publication Date
2021-07-15
Journal
Publisher
ISSN
Cited by
5469
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest
References available in scid.ai4
MAFFT Multiple Sequence Alignment Software Version 7: Improvements in Performance and Usability2013
Features and development of Coot2010
UCSF ChimeraX : Structure visualization for researchers, educators, and developers2020
Macromolecular structure determination using X-rays, neutrons and electrons: recent developments in Phenix2019
Cited by11
Accurate structure prediction of biomolecular interactions with AlphaFold 32024
Protein complex prediction with AlphaFold-Multimer2021
Computational approaches streamlining drug discovery2023
Computer-Aided Drug Design and Drug Discovery: A Prospective Analysis2023
Scalable emulation of protein equilibrium ensembles with generative deep learning2025
Empowering biomedical discovery with AI agents2024
Data‐Driven Protein Engineering for Improving Catalytic Activity and Selectivity2023
Klebsiella phage KP34gp57 capsular depolymerase structure and function: from a serendipitous finding to the design of active mini-enzymes against K. pneumoniae2023
Protein language models trained on biophysical dynamics inform mutation effects2026
Investigating the determinants of performance in machine learning for protein fitness prediction2025
Catalytically active inclusion bodies of Bacillus clausii laccase protein recombinantly produced in E. coli for dye decolorization2026