Proteus: exploring protein structure generation for enhanced designability and efficiency

The development of de novo protein design method is crucial for widespread applications in biology and chemistry. Protein backbone diffusion aims to generate designable protein structures with high efficiency. Although there have been great strides in protein structure prediction, applying these met...

Full description

Saved in:
Bibliographic Details
Published inbioRxiv
Main Authors Wang, Chentong, Qu, Yannan, Peng, Zhangzhi, Wang, Yukai, Zhu, Hongli, Chen, Dachuan, Cao, Longxing
Format Paper
LanguageEnglish
Published Cold Spring Harbor Cold Spring Harbor Laboratory Press 15.02.2024
Subjects
Computer applications
Diffusion
Efficiency
Protein structure
Proteins
Online AccessGet full text

Cover

More Information
Summary:The development of de novo protein design method is crucial for widespread applications in biology and chemistry. Protein backbone diffusion aims to generate designable protein structures with high efficiency. Although there have been great strides in protein structure prediction, applying these methods to protein diffusion has been challenging and inefficient. We introduce Proteus, an innovative approach that uses graph-based triangle methods and a multi-track interaction network. Proteus demonstrated cutting-edge designability and efficiency in computational evaluations. We tested the reliability of the model by experimental characterization. Our analysis indicates that from both computational and experimental perspectives, it is able to generate proteins with a remarkably high success rate. We believe Proteus's capacity to rapidly create highly designable protein backbone without the need for pre-training techniques will greatly enhance our understanding of protein structure diffusion and contribute to advances in protein design.Competing Interest StatementThe authors have declared no competing interest.Footnotes* paper format modified
DOI:10.1101/2024.02.10.579791