S. Ovchinnikov

Predicting absolute protein folding stability using generative models

The paper uses generative models to predict absolute protein stability, achieving a mean error of 1.5 kcal/mol and a correlation coefficient of 0.7.

Cyclic peptide structure prediction and design using AlphaFold

This paper modifies AlphaFold to predict and design cyclic peptides, achieving high accuracy in structure prediction and de novo design with experimental validation.

Mega-scale experimental analysis of protein folding stability in biology and protein design

The paper introduces cDNA display proteolysis to measure protein folding stability for thousands of protein domains, revealing environmental factors, thermodynamic couplings, and evolutionary divergence in amino acid usage.

Prediction of multiple conformational states by combining sequence clustering with AlphaFold2

The paper uses sequence clustering with AlphaFold2 to predict conformational states of proteins and identifies mutations that switch between states.

State-of-the-Art Estimation of Protein Model Accuracy using AlphaFold

This paper shows AlphaFold has learned a biophysical energy function to predict protein structures from sequences, achieving state-of-the-art accuracy without coevolution data.

Deep learning methods for designing proteins scaffolding functional sites

This paper introduces two deep learning methods for designing proteins with specific functional sites, showing they can generate diverse proteins with desired properties.

A structural biology community assessment of AlphaFold 2 applications

This study assesses AlphaFold 2's applications in structural biology, finding it outperforms existing tools in predicting disorder, complexes, and structural features with high confidence.

BacPROTACs mediate targeted protein degradation in bacteria

This paper develops BacPROTACs, small-molecule degraders that target bacterial proteins by activating the ClpC:ClpP protease, demonstrating in vivo degradation in mycobacteria.

Design of proteins presenting discontinuous functional sites using deep learning

This paper designs proteins with discontinuous functional sites using deep learning, eliminating the need for pre-defined secondary structures and enabling diverse scaffold generation.

Protein sequence design by explicit energy landscape optimization

This paper designs protein sequences by optimizing an energy landscape to ensure they fold into a desired structure, showing improved performance over traditional methods.

De novo protein design by deep network hallucination

The paper uses deep networks to design new proteins by inverting structure prediction models, resulting in 27 stable folded proteins with diverse structures.

An Empirical Demonstration of Unsupervised Machine Learning in Species Delimitation

This study demonstrates that unsupervised machine learning methods can effectively identify species-level divergences in genetic data, outperforming traditional model-based approaches in avoiding over-splitting species.