protein sequence

The paper uses deep learning to show estradiol binds androgen receptors more strongly than testosterone, revealing evolutionary conservation and potential therapeutic implications.

This study shows that mitochondrial presequences function in cytosolic quality control and trigger a heat shock response when mitochondrial import is inhibited.

The paper identifies conserved influenza A epitope regions and evaluates ESM-2 sequence features for epitope prioritization, finding no significant improvement over alignment-based conservation.

ICFinder uses protein language models to identify ion channels and predict permeation residues, outperforming existing methods and providing a public resource.

TRACER predicts terpene skeletal rearrangements by learning enzyme-product associations, identifying novel carbon skeletons and mechanistic insights.

SAMP V2 improves antimicrobial peptide identification by integrating biochemical and sequence-order information into split amino acid composition, outperforming existing models on multiple datasets.

ESM-ECForest uses ESM-2 embeddings and Random Forest classifiers to predict enzyme functions with high accuracy across EC levels.

This study introduces CrossVOI, a framework for VOC-OR interaction prediction using protein language models and cross-attention, showing improved performance and interpretability of attention mechanisms in human olfactory receptors.

The paper benchmarks ESM-2 model precision configurations for variant-effect prediction, finding that average performance metrics hide critical failures and that asymmetric activation quantization improves accuracy.

The paper uses computational methods to redesign β-lactoglobulin, improving its stability and BCAA content, with experimental validation of the results.

FlyPredictome scores 1.5 million AlphaFold-Multimer predictions of Drosophila protein pairs with a contact-aware integrated Local Interaction Score (iLIS), yielding residue-level interfaces enriched for phenotype-associated missense alleles.

SilkRoute provides a reproducible framework for constructing multi-source biomolecular datasets through descriptor-driven workflows.

This paper develops interpretable biophysical neural networks to predict transcriptional activation domains from protein sequences, revealing how residue properties influence both activation strength and protein abundance.

This paper aligns a protein generative model with experimental fitness using Direct Preference Optimization, improving stability prediction and enabling new applications.

ProtPen combines sequence and structure-based methods to annotate proteins of unknown function across proteomes with >90% accuracy.

The paper develops a pathogenicity prediction model for hemoglobin variants using structure-informed features and ML, showing strong performance compared to AlphaMissense and providing a complete mutational landscape for HbA.

This study evaluates how large language models perform in-class learning for antibody classification using biological sequences, finding that few-shot prompting with diverse demonstrations achieves competitive results.

This paper shows that focusing on complementarity-determining regions (CDRs) in antibody sequences improves binding affinity prediction using region-aware pretraining strategies.

This paper introduces a framework for comparing embedding spaces in molecular biology, demonstrating its utility in pre-filtering machine learning tasks and identifying differential patterns across embeddings.

The paper examines subclonal complete loss of CDKN1B in prostate cancer, finding higher prevalence in African American individuals and associations with worse outcomes.

This paper introduces vhPPIpred, a machine learning method for predicting virus-human PPIs that outperforms existing methods and provides a benchmark dataset.

A fine-tuned ProtT5 protein language model predicts per-residue energetic frustration from sequence (Pearson r 0.86 on unseen CATH topologies) and profiles the human proteome in ~17 minutes on one GPU.

REvolutionH-tl automates evolutionary gene history analysis from sequence data, outperforming existing tools in speed and accuracy with integrated visualization.

Pro4S predicts protein solubility by integrating sequence, structure, and surface data, improving accuracy and reducing non-expressed proteins in de novo design.

The paper introduces a framework for multi-type PTM site prediction using protein language model embeddings, achieving improved performance in a shared-model setting.

ESMDynamic predicts protein dynamic contact maps from sequences, outperforming existing methods with less computation and enabling large-scale analysis of conformational variability.

This study fine-tunes a pre-trained language model to predict antibody hydrophobicity using retention time data, achieving high accuracy with rigorous data quality control.

Open plastid orthogroup resource for algae: 496 named orthogroups from 893 genomes, 224 core profile hidden Markov models with calibrated cutoffs, alignments, predicted structures, 859-genome phylogeny, CheckM-compatible marker sets.

The paper evaluates how data preprocessing steps affect protein function prediction performance using antioxidant protein classification.

AbSLang-search efficiently retrieves structurally similar antibodies from large sequence databases using a contrastive learning model for CDR RMSD prediction.