Andrew M. Watkins

DyAb predicts antibody properties and designs new sequences with high binding affinity using a low-data approach.

ConformAb designs antibody sequences that maintain CDR canonical conformations, improving binding affinity without target-specific data.

Property Enhancer optimizes antibody properties using a data-efficient framework, achieving significant affinity gains in both in silico and in vitro tests.

This paper develops a machine learning model to predict Trp oxidation risk in antibodies and shows that modulating electrostatic potential reduces oxidation while maintaining binding affinity.

BRIDGE is a graph neural network that embeds protein-protein interactions for therapeutic applications, achieving state-of-the-art performance in predicting Ab-Ag affinity and antibody viscosity.

This paper introduces a machine learning framework for therapeutic antibody design that integrates computational methods with experimental validation, resulting in significantly improved binding affinity for multiple targets.

The paper trains a new version of AlphaFold2, OpenFold, and evaluates its generalization ability and learning mechanisms.

The paper uses cryo-EM to study T-box-tRNA interactions and shows how structural modifications enhance selectivity and sensitivity in translational regulation.

This study uses community science and experimental screening to design mutant ribosomes with improved protein synthesis and cell growth in vitro and vivo.

EquiFold predicts all-atom protein structures using an SE(3)-equivariant coarse-grained geometric representation without sequence alignments or language model embeddings, matching AlphaFold accuracy orders of magnitude faster in preliminary tests.

The paper designs and selects ribosomal active site mutants with high activity using a computational approach to overcome limitations in rRNA modification.

This study develops PERSIST-seq and In-line-seq to optimize mRNA stability and translation, finding that in-cell stability is more critical than ribosome load for protein output.

This paper introduces a deep learning method for protein design that uses a sequence denoising autoencoder and function predictor to guide sampling towards functional sequences.

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.

The paper develops and validates rotamer libraries for beta3-amino acids to support their inclusion in protein design frameworks.