Matt McPartlon

Drug-like antibody design against challenging targets with atomic precision

This paper introduces a computational antibody design platform that generates drug-like antibodies with high developability and atomic accuracy, demonstrating its utility for challenging therapeutic targets.

Zero-shot antibody design in a 24-well plate

This paper introduces Chai-2, a model that designs antibodies with a 16% hit rate, achieving success in 50% of targets without prior binders, and enables rapid validation in under two weeks.

Chai-1: Decoding the molecular interactions of life

Chai-1 matches AlphaFold3 on PoseBusters ligand poses (77% vs 76%), beats AlphaFold-Multimer 2.3 on low-homology interfaces, and stays competitive without MSAs on multimers and antibodies.

PINDER: The protein interaction dataset and evaluation resource

PINDER provides a benchmark dataset for evaluating protein-protein interaction docking models, revealing overestimation of performance on leaky test sets and showing improvements with interface-clustered training.

Deep Learning for Flexible and Site-Specific Protein Docking and Design

This paper introduces DockGPT, a deep learning method for protein docking that improves accuracy and speed, especially for antibody-antigen complexes, and extends to co-designing antibody sequences and structures.

Unlocking de novo antibody design with generative artificial intelligence

Deep generative models combined with high-throughput screening design de novo anti-HER2 antibodies without further optimization, yielding nanomolar binders functionally comparable or superior to trastuzumab.

End-to-End deep structure generative model for protein design

This paper introduces a VAE-based generative model for protein design that generates high-quality 3D structures and outperforms existing methods in sequence design.

A Deep SE(3)-Equivariant Model for Learning Inverse Protein Folding

This paper introduces a deep SE(3)-equivariant model for inverse protein folding, achieving state-of-the-art results in sequence recovery and mutation effect prediction.

AttnPacker: An end-to-end deep learning method for rotamer-free protein side-chain packing

AttnPacker predicts protein side-chain conformations with lower RMSD than existing methods, achieving top performance in dihedral prediction on CASP13 and CASP14 data.

Improved protein structure prediction by deep learning irrespective of co-evolution information

This study shows that deep learning models like ResNet can predict protein structures accurately without co-evolution data, outperforming existing methods on CASP13 targets and human-designed proteins.