Protein Structure and Dynamics

HyRes simulates dynamic protein structures and interactions for IDPs, showing high accuracy across environments and generating ensembles for thousands of human IDPs.

This study uses deep learning to redesign an antifreeze protein, identifying variants with improved properties but not surpassing wild-type activity.

ComplexDesign is a method for designing multichain protein complexes by optimizing sequences to fold proteins and form inter-chain interactions, achieving high success rates in designing dimers, trimers, and tetramers.

The paper investigates the energy landscape of coupled folding and binding in intrinsically disordered proteins, finding that electrostatic interactions compensate for energetic frustration.

The paper introduces CNR, a framework to quantify how protein contact networks respond to stabilizing interaction changes, using human AlphaFold models to analyze 22,167 structures.

The paper introduces a docking-based simulation method that combines rigid-body approximations with flexible protein structures to model macromolecular interactions in crowded cellular environments over long timescales.

kontakteUR transforms structural coordinates into a chemically informed contact space to analyze biomolecular interactions and their changes, enabling systematic comparison and functional insights.

The paper uses molecular dynamics simulations to show that phosphate release from actin filaments is limited by dissociation from Mg2+ and varies with subunit position and water molecule count.

SteerAF improves AlphaFold2's ability to predict alternative protein conformations by using distogram information from MSAs, achieving better results than existing methods.

The paper shows that a spontaneous post-translational succinimide modification in a bacterial solute-binding protein enhances its thermostability and binding affinity for 5-oxoproline.

The paper identifies Ki-67-like proteins in non-vertebrate eukaryotes, showing conserved biophysical and functional properties despite low sequence homology.

This paper introduces dEVA, a framework for de novo metalloenzyme design, achieving high catalytic activity without natural templates.

PolyFold evaluates structure prediction models by filtering homology with a new method, showing reduced performance gaps between Boltz-2 and OpenFold 3.

This paper introduces HomoDSP, a method for DNA-binding specificity prediction using homologous protein-DNA structures, showing improved accuracy and generalization over existing methods.

The paper identifies a histidine switch that regulates pH-responsive self-assembly of a helical protein filament through electrostatic interactions.

The paper shows that machine learning models and large datasets fail to accurately predict extremophile enzyme melting and optimum temperatures, due to biased and error-laden training data.

The paper uses molecular dynamics simulations to identify a unifying mechanism for excipient-mediated protein thermal stability based on shape complementarity between proteins and solvents.

The paper identifies transposable elements as a source of intrinsic disorder in human proteins, showing that TE insertions are enriched in disordered regions and their conformational fate depends on their age and TE class.

The paper evaluates how protein fitness models predict and design variants with altered specificity, finding that models biased against such changes can be leveraged by using weighted differences in model scores.

The paper tests AlphaFold2 and ColabFold's reliability by using false protein sequences and finds a low false positive rate, while also discovering new pseudogenes in the human genome.

This study investigates the membrane-dependent structure of CPXV012 and its interaction with TAP, revealing a two-helical organization and flexible ER-luminal region critical for TAP recognition.

This study shows that knotted protein structures inhibit aggregation by limiting expanded states and suppressing liquid-liquid phase separation.

This paper introduces a mathematical framework to analyze allostery in proteins, showing how mutations affect pathway correlations and communication through intermediary residues.

The paper investigates how amino acid homorepeats (HRs) in proteins influence molecular outcomes through different mosaic configurations and identifies their functional implications.

ProMiSE evaluates protein multi-state prediction models using a curated dataset of biological conformational changes, revealing limitations in capturing functionally relevant dynamics.

The paper develops a structure prediction method using XL-MS data for disordered microsporidian proteins with no homologs, successfully modeling two spore wall proteins.

MAHLER predicts antibody-antigen dissociation kinetics using a hybrid machine learning and physics approach, achieving screening-grade accuracy and fast computation.

The paper validates MHPC512, a special-purpose supercomputer for molecular dynamics simulations, showing it produces results consistent with established software.

EditorForge refines inverse-folded protein sequences for enzyme and genome-editor domains by applying constraints and audits to ensure structural integrity and active-site protection.

This paper shows that arginine substitutions, despite similar charge, compact α-synuclein more than lysine, accelerating amyloid formation and reshaping its disordered ensemble.