protein structure prediction

The paper presents the first high-resolution structure of a conjugative pilus tip from the R388 T4SS, revealing a structural mechanism that couples tip contact to channel opening and identifies convergent principles in bacterial secretion systems.

The paper introduces coarse-tuning to specialize a protein foundation model for a specific domain, showing improved performance on transfer learning tasks using a medium-sized dataset of outer membrane beta barrels.

This study determines the structure of full-length human hephaestin using cryo-EM, revealing its homodimeric organization and substrate-binding site.

RNAtomy is a structure-aware RNA foundation model that learns structural and functional representations from geometric and conformational features, achieving strong performance on diverse tasks.

This paper designs de novo ligases that achieve site-specific ubiquitylation of a target protein with high selectivity and efficiency.

The paper audits protein generate-and-screen pipelines by testing steering, confidence, and activation readouts, finding they do not reliably predict foldability.

The paper introduces a protein language model pretrained on continuous biophysical fields, showing improved performance on thermostability prediction compared to sequence-only models.

DiffTernary is a diffusion model for predicting molecular glue ternary complex structures, achieving competitive performance on benchmark datasets.

The paper extends a framework to generate atomic resolution ensembles of intrinsically disordered proteins using six AI structure predictors and compares their performance against experimental data and simulations.

This study determines the cryo-EM structures of human claudin-18.2 bound to therapeutic antibodies, revealing key residues involved in antibody recognition.

REVEAL captures RNA excited states using AI-guided simulations, showing it can identify dominant conformations better than traditional methods.

DaliScope analyzes DALI structural alignments to uncover protein relationships and structural patterns, revealing unannotated domains and structurally anchored sequence signatures.

The paper characterizes LOca-1, a calcium-activated arginyl-specific orthocaspase from Loki's Castle, showing its thermoactivity and structural conservation with plant metacaspases.

The paper evaluates AlphaFold's predictions of transmembrane protein complexes, finding that while most show plausible membrane geometry, some have inconsistencies and limitations in predicting lipid interactions.

OmniT2I converts molecular profiles into images and predicts survival and missing modalities from incomplete cancer data, achieving higher accuracy than separate models.

The paper introduces WhatRes, a tool to evaluate AI-predicted protein structures by comparing their geometry to experimental structures across resolution and realism.

The paper identifies bacterial MVP homologs and reports cryoEM structures of vaults from Corallococcus praedator and a Deltaproteobacteria bacterium, revealing novel symmetries.

The paper predicts a cohesin acetylation complex in budding yeast that interacts with nicked DNA and confirms its function through experiments.

StableProt predicts protein thermostability and organismal optimal growth temperature using a structure-aware deep learning framework, achieving high accuracy and robustness across temperature regimes.

The paper reports the cryo-EM structures of an Acinetobacter baumannii podovirus, revealing its capsid maturation mechanism and providing a reference for phage engineering.

This paper evaluates generative models for predicting novel enzymatic reactions, finding they can generate correct reactions for 25% of test cases and improve over existing methods.

RPFace predicts RNA-protein interfaces directly from sequence, outperforming existing methods on RNA interfaces and generalizing to in vivo data without pretraining.

This paper introduces PDLMs, protein-DNA language models that jointly represent protein and DNA sequences to predict and engineer DNA recognition specificity.

RPI-Zero predicts RNA-protein interactions without prior data by using a sequence-only model trained on a large benchmark of CLIP experiments.

This study uses molecular dynamics simulations to show that the π-conformation of PKD2 supports ion conduction, challenging the previous model that the α-conformation is the conductive state.

The paper identifies novel Cas9 orthologs from metagenomic data and develops NanoPAMDA to rapidly characterize their PAM preferences.

PALM, a protein language model, predicts domain, functional site, and variant effects with high accuracy using curated sequence data and outperforms existing models in multiple benchmarks.

This paper introduces CAISL, a framework for improving survival prediction in lung adenocarcinoma by integrating clinical and molecular data with stability constraints, achieving better performance than clinical and molecular-only models.

FoldAbS repurposes protein folding models as foundation encoders to improve antibody screening by evaluating specificity, affinity, and developability in a unified framework.

This study evaluates AlphaFold 3's accuracy and reliability in predicting RNA structures and protein-RNA interfaces, finding that confidence scores do not always correlate with correct predictions.