Jeffrey J. Gray

This paper evaluates AlphaFold3's ability to predict supramolecular self-assembly of peptides, finding it captures trends but has limitations in modeling complex assemblies and charge effects.

DFMDock is a diffusion model for protein docking that combines generative sampling and energy-based ranking, achieving higher success rates than existing methods.

This paper adapts co-folding models for structure-based protein-protein docking by replacing their template module with a docking module and training with a flow-matching objective, showing competitive performance on specific targets.

ProtAff predicts protein-protein binding affinity using sequence data, outperforming existing methods on an EGFR benchmark but underperforming structural methods in cross-target discrimination.

This paper introduces GermRL, a reinforcement learning framework that reduces germline bias in antibody language models, improving mutation diversity while maintaining structural plausibility.

The paper introduces two models, PiCAP and CAPSIF2, to predict protein-carbohydrate interactions at the protein and residue levels, achieving high accuracy and demonstrating substantial interplay across proteomes.

This paper benchmarks AI models on predicting antibody developability properties using FLAb2, finding they cannot consistently predict most properties and performance depends on data and architecture.

This paper examines whether diffusion models can extract physics-like energies for protein docking, finding they can recover energy funnels comparable to Rosetta in 9 out of 25 cases.

The paper evaluates deep learning models for predicting noncovalent protein-carbohydrate interactions using a new dataset and metric, finding comparable performance but declining accuracy with longer carbohydrates.

This study evaluates AlphaFold3's ability to predict antibody and nanobody structures and their docking with antigens, finding high accuracy but significant room for improvement.

This study shows that anti-citrullinated protein antibodies (ACPAs) in RA arise from germline antibodies targeting carbamylated proteins through affinity maturation and somatic hypermutations.

The paper introduces a method to learn protein and antibody residue representations using equivariant graph transformers, showing structural context can determine amino acid labels and improve antibody design.

The paper develops a computational method to design diverse venom libraries and discovers peptides targeting the human itch receptor MRGPRX4.

This paper introduces Rosetta-MPDock, a membrane protein docking method that improves prediction accuracy by capturing conformational changes and integrating AlphaFold2-multimer, achieving better success rates in predicting membrane protein complex structures.

The paper identifies a single-subunit oligosaccharyltransferase from Desulfovibrio marinus that enables glycosylation of full-length IgG antibodies in E. coli.

This study uses DeepAb to design antibody variants with enhanced stability and affinity without antigen information, showing 91% and 94% improvements in thermal/colloidal stability and affinity, respectively.

FLAb benchmarks deep learning models for predicting antibody fitness properties across six metrics, finding no model consistently correlates with all properties.

GeoDock predicts protein-protein docked structures using a multi-track iterative transformer, achieving high success rates and fast inference speeds for rigid and flexible targets.

This paper develops machine learning models to predict antibody thermostability using sequence and structure features, showing better generalization with a CNN over pre-trained language models.

The paper introduces FvHallucinator, a deep learning model that generates antibody CDR libraries conditioned on structure, improving binding affinity and enriching for therapeutic candidates.

The paper introduces ReplicaDock 2.0, a method that improves protein complex structure prediction by simulating induced-fit dynamics, achieving better accuracy for flexible targets.

The paper introduces IgFold, a deep learning method for fast and accurate antibody structure prediction, achieving results comparable to AlphaFold in under a minute.

The paper presents a framework for ensuring reproducibility in bio-macromolecular modeling through automated benchmarks and a test server, demonstrating its utility for Rosetta and other tools.

This paper expands the Rosetta software suite with tools for glycan modeling, density fitting, and rational sequon design, achieving high accuracy in predicting and building glycan structures.

This paper introduces a method to construct and analyze neoglycoproteins by systematically modifying glycosylation sites in proteins and assessing the effects of glycan addition.