Related Experiment Video
Updated: Jul 16, 2025

Optimization of Synthetic Proteins: Identification of Interpositional Dependencies Indicating Structurally and/or Functionally Linked Residues
Published on: July 14, 2015
A probabilistic view of protein stability, conformational specificity, and design.
Jacob A Stern1, Tyler J Free2, Kimberlee L Stern3
1Department of Computer Science, Brigham Young University, Provo, UT, USA.
This study introduces BayesDesign, a novel algorithm for protein sequence design that optimizes protein stability and specificity. It outperforms existing methods in enhancing protein properties, offering a new approach to inverse folding.
Area of Science:
- Computational biology
- Protein engineering
- Machine learning for protein design
Background:
- Neural networks are used as probabilistic models for protein sequence design.
- Existing inverse folding models use various objective functions with unassessed trade-offs.
Purpose of the Study:
- To introduce probabilistic definitions of protein stability and conformational specificity.
- To demonstrate the link between these properties and the Boltzmann probability objective.
- To propose and evaluate a novel sequence decoding algorithm, BayesDesign.
Main Methods:
- Probabilistic definitions of protein stability and conformational specificity were developed.
- The relationship between these properties and the Boltzmann probability objective was demonstrated.
- A novel algorithm, BayesDesign, was proposed, utilizing Bayes' Rule to maximize the Boltzmann probability objective.
Main Results:
- BayesDesign successfully links the Boltzmann probability objective to experimentally verifiable outcomes.
- Both BayesDesign and ProteinMPNN enhanced the thermostability of the NanoLuc enzyme.
- Both algorithms increased the conformational specificity of the WW structural motif.
Conclusions:
- BayesDesign offers a new, effective approach to inverse folding by optimizing the Boltzmann probability objective.
- The study validates the link between probabilistic definitions and experimental outcomes in protein design.
- Further analysis of potential model error sources is provided.
Related Concept Videos
Protein Folding
Protein Denaturation
Conservation of Protein Domains Over Different Proteins
A limited set of protein domains often duplicate and recombine during evolution. These domains can be organized in different combinations to...
Intrinsically Disordered Proteins
Protein Organization
The primary structure of a protein is its amino acid sequence....
Conserved Binding Sites
Binding sites are often located in large pockets, and if their location on a protein’s surface is unknown, it can be predicted using various approaches. The energetic method computationally...

