Related Experiment Video
Updated: Jun 18, 2025
09:47
Author Spotlight: Advancing Alzheimer's Research – Exploring Early Detection and Multi-Omics Approaches
Published on: December 15, 2023
1.0K
Parametrically guided design of beta barrels and transmembrane nanopores using deep learning
David E Kim1,2,3, Joseph L Watson1,2,3, David Juergens1,2,3
1Department of Biochemistry, University of Washington, Seattle, WA 98195.
Biorxiv : the Preprint Server for Biology
|August 2, 2024
Summary
Parametric design of beta barrels is now possible using deep learning, enabling precise control over protein shape and function. This method successfully created novel transmembrane nanopores with high accuracy and experimental success rates.
Area of Science:
- Protein design
- Structural biology
- Computational biophysics
Background:
- Global parameterization successfully guided coiled coil protein design.
- Beta barrel design using 2D blueprints is complex and offers indirect shape control.
- Deviations from ideal geometry hinder beta barrel design.
Purpose of the Study:
- Generalize parametric design to beta barrels.
- Integrate parametric methods with deep learning for protein design.
- Develop a more accessible and controllable method for designing beta barrel proteins.
Main Methods:
- Utilized RoseTTAFold-based methods (RFjoint inpainting and RFdiffusion).
- Started with parametrically generated barrel backbones.
- Incorporated backbone irregularities for proper folding.
Main Results:
- Achieved high in silico and experimental success rates for beta barrel design.
- Confirmed atomic accuracy with X-ray crystallography of a novel topology.
- De novo designed transmembrane nanopores (12, 14, 16-stranded) with conductances of 200-500 pS.
Conclusions:
- Parametric generation combined with deep learning enhances beta barrel design.
- The new approach offers simplicity, control, and high success rates.
- Enables precise specification and accessibility for designing functional beta barrel proteins like nanopores.
