Protein Structure and Dynamics

Advancing Protein Design via Multi-Agent Reinforcement Learning with Pareto-Based Collaborative Optimization

MAProt improves protein design by combining structure-based and language model approaches using multi-agent reinforcement learning with Pareto-based optimization, achieving better performance across multiple tasks.

ProAR: Probabilistic Autoregressive Modeling for Molecular Dynamics

ProAR generates molecular dynamics trajectories with reduced error and improved conformational accuracy compared to existing methods.

An essential dynamics-based elastic network model to unravel the conformational dynamics of DNA, RNA, and protein-nucleic acid complexes

The paper introduces edENM, an elastic network model refined for DNA, RNA, and protein-nucleic acid complexes, showing better agreement with experimental data than previous models.

Multiscale conformational sampling of multidomain fusion proteins by a physics informed diffusion model

This paper introduces a physics-informed diffusion model to efficiently sample conformational ensembles of multidomain fusion proteins, outperforming traditional methods in capturing large-scale dynamics.

Enhanced Sampling on Domain/Motif Level with Kinetic Accelerated Molecular Dynamics

This paper introduces KAMD, an enhanced sampling method that improves molecular dynamics efficiency for domain/motif level processes while maintaining equilibrium properties.

Do AI Models for Protein Structure Prediction Get Electrostatics Right?

This study evaluates AI models for protein structure prediction by testing their ability to correctly place ionizable residues, finding they often bury these residues in non-polar regions despite physico-chemical principles.

pH-dependent structural dynamics of the neuropeptide Y in aqueous solution

This study investigates the pH-dependent structural dynamics of neuropeptide Y using NMR and constant pH simulations, revealing pH-induced changes in helical core length and protonation states.

Generalise or Memorise? Benchmarking Ligand-Conditioned Protein Generation from Sequence-Only Data

This study benchmarks ligand-conditioned protein generation using sequence-only data, revealing trade-offs between diversity and foldability based on dataset characteristics.

Counting strands in outer membrane beta-barrels

The paper introduces an algorithm to accurately count strands in bacterial outer membrane beta-barrels, achieving 97% accuracy on 571,760 structures from AlphaFold2.

Determinants of Improved CGRP Peptide Binding Kinetics Revealed by Enhanced Molecular Simulations

This study identifies molecular determinants of improved CGRP peptide binding kinetics using enhanced simulations of the CLR:RAMP1 complex and its engineered variant ssCGRP.

Solution Structure of the Novel CH-domain zinc finger from the puberty regulator Makorin-3

This paper determines the solution structure of the CH-domain zinc finger from Makorin-3, revealing its metal-binding properties and potential role in protein interactions.

ProtNHF: Neural Hamiltonian Flows for Controllable Protein Sequence Generation

ProtNHF enables controllable protein sequence generation through Hamiltonian flows with inference-time bias functions, achieving competitive quality and flexible property control.

Rigidity-Aware Geometric Pretraining for Protein Design and Conformational Ensembles

This paper introduces RigidSSL, a geometric pretraining framework for protein design and conformational modeling, improving designability and capturing realistic dynamics.

Characterizing the Conformational Dynamics of an Intrinsically Disordered Localization Sequence

This study examines how single-residue substitutions in a mitochondrial localization peptide affect its conformational dynamics, revealing that sequence changes can bias structural ensembles and influence targeting efficiency.

Molecular mechanisms of dynamics-mediated effects of pathogenic missense mutations on Menin protein function

This study shows that pathogenic missense mutations affect Menin function through dynamic changes and allostery, not just structural destabilization.

From Prefix to Path: Learning Temporally Consistent Biomolecular Dynamics from Limited Initial Data

This paper introduces a Transformer-based framework to predict full molecular dynamics trajectories from limited initial data, showing it captures long-range temporal structures and improves sampling compared to traditional methods.

Confidence Without Verification: Screening pLDDT Unreliability in AlphaFold2 Fold-Switching Predictions

The paper identifies unreliable confidence metrics in AlphaFold2 predictions for fold-switching proteins and introduces a screening method to detect false positives.

Direct evidence of acid-driven protein desolvation

The paper examines hydration changes in a protein complex across pH values using cryo-EM and simulations, showing significant water loss and its link to metal release.

An Energy Landscape Approach to Miniaturizing Enzymes using Protein Language Model Embeddings

The paper uses a PLM-guided energy landscape approach to design compact enzymes with preserved catalytic site structures, validated through structure prediction and molecular dynamics simulations.

Progressive Backmapping of Highly Coarse-Grained Protein Models

This paper introduces a progressive backmapping framework to reconstruct all-atom protein structures from highly coarse-grained models, achieving high accuracy across multiple resolutions and complex assemblies.

Structural Plausibility Without Binding Specificity: Limits of AI-Based Antibody-Antigen Structure Prediction Confidence Scores

The paper evaluates AI-based antibody-antigen structure prediction methods, finding that confidence scores often fail to distinguish correct from incorrect pairings, even with increased sampling.

Density-guided AlphaFold3 uncovers unmodelled conformations in β2-microglobulin

This study uses density-guided AlphaFold3 to reveal unmodeled conformational heterogeneity in β2-microglobulin from crystallographic maps.

Dynamical buffering of reconfiguration dynamics in intrinsically disordered proteins

The paper investigates how sequence variation affects the dynamics of intrinsically disordered proteins, finding that reconfiguration times are largely independent of sequence despite changes in chain dimensions.

ProPrep: An Interactive and Instructional Interface for Proper Protein Preparation with AMBER

ProPrep streamlines molecular dynamics preparation by guiding users through steps with automation and visualization, enabling reproducible and transparent simulations.

High Diversity Gene Libraries Facilitate Machine Learning Guided Exploration of Fluorescent Protein Sequence Space

This study shows that expanding training data diversity with synthetic gene libraries improves machine learning models' ability to generate functional fluorescent proteins beyond natural sequence space.

intDesc-AbMut: Describing and understanding how antibody mutations impact their environmental interactions

The paper introduces intDesc-AbMut, a tool to automate the analysis of antibody mutations' effects on environmental interactions, achieving high accuracy in predicting crystal-like environments.

A Systematic Evaluation of Protein Phase Separation Predictors Across Diverse Protein Landscapes

The paper evaluates nine LLPS predictors across diverse protein landscapes, finding significant variability in performance and highlighting the need for diverse benchmarking data.

Adversarial Sequence Mutations in AlphaFold andESMFold Reveal Nonphysical StructuralInvariance, Confidence Failures, and Concerns forProtein Design

The paper evaluates AlphaFold and ESMFold by introducing adversarial mutations and finds structural invariance and unreliable confidence metrics, suggesting reliance on templates rather than biophysical principles.

Hybrid Deep Learning with Protein Language Models and Dual-Path Architecture for Predicting IDP Functions

IDPFunNet predicts IDR functions using a hybrid deep learning model, outperforming existing methods in binding prediction and DFL identification across multiple benchmarks.

Evolutionary constraints improve protein large language model predictions for protein stability, binding regions and epistasis

D2D improves protein mutation effect predictions using evolutionary constraints and a language model, outperforming existing methods in stability, binding, and epistasis.