protein sequence

The paper identifies and classifies ion channels across species, revealing functional insights into understudied CALHM channels through sequence, structure, and functional data integration.

TreeGazer selects informative protein sequences using phylogenetic structure and Bayesian optimization, outperforming alternative strategies in simulated campaigns and low-data settings.

This paper introduces AminoX, a platform for programmable covalent protein design using non-standard amino acids, enabling efficient screening and functional validation of covalent warheads and multifunctional proteins.

This paper improves variant effect prediction by modifying protein language models with sparse mechanistic features, achieving significant performance gains across multiple datasets.

Bio-BLIP is a multimodal architecture that integrates DNA, genes, proteins, and text to improve variant interpretation without task-specific fine-tuning, achieving better performance in variant prioritization and target gene prediction.

APEX identifies pathogen targets and designs inhibitors using explainable AI, highlighting key residues and guiding molecular design.

The paper introduces Aiki-Sol, a predictor for E. coli protein solubility that accounts for centrifugation regimes, improving performance by treating regime as a feature rather than noise.

The paper shows that structure tokens in ESM-3 enhance the shared biological vocabulary of protein language models without creating a new feature set.

The paper introduces a method to select Reference Proteomes for UniProt, increasing coverage and reducing redundancy in the knowledgebase.

This paper presents a deep learning framework for protein-ligand binding prediction using BiLSTM and GCN, achieving high AUROC and AUPRC without 3D structural data.

This study shows SUMO1 and SUMO2 differentially regulate TDP-43 phase separation and aggregation, with SUMO1 inhibiting and SUMO2 promoting liquid-like states that prevent aggregation.

The paper identifies ABHD13 and ABHD16A as deacylases regulating SNAP25 family proteins and maps substrate selectivity to a single amino acid position.

This paper develops an uncertainty-aware framework for PAD biomarker discovery using graph neural networks and positive-unlabeled learning, identifying novel and validated candidate proteins.

The paper fine-tunes a protein language model for PTM-site prediction and shows how model size, annotation depth, and modification chemistry affect PTM recognition.

The paper benchmarks sequence- and structure-based protein models using a unified framework, showing structure-based models outperform others in generating functional variants.

BindPred predicts protein-protein binding affinity from amino acid sequences using language model embeddings and gradient boosting trees, achieving high correlation on a benchmark dataset.

BAT constructs and annotates gene trees to explore functional relationships within gene families using local genomic resources.

LysinFusion improves phage lysin prediction using a hybrid CNN-Transformer model, achieving higher accuracy and fewer false positives than existing methods.

ProteinFlux predicts protein conformational dynamics influenced by post-translational modifications using a generative framework, outperforming traditional methods in speed and accuracy.

This study uses generative protein design to explore how sequences encode amyloid fibril structures, showing a continuous compatibility between sequences and a common amyloid fold.

PRISM improves antibody language modeling by distinguishing germline and non-germline residues, achieving better binding affinity correlation and controllable generation.

This paper develops two codon optimization tools using classical and AI methods, showing they outperform existing tools in protein expression.

This paper introduces a deep learning framework for designing proteins with site-specific post-translational modifications using latent diffusion models and a PTM-aware language model.

The paper develops a two-step in vitro evolution method to improve translational proteins and identifies mutations that enhance translation activity.

This study compares deep learning models for generating antimicrobial peptides, finding that auto-encoders, especially Wasserstein auto-encoders, produce more diverse sequences than other models.

The paper audits structural biases in ML models for antimicrobial peptide design and finds that sequential training data lead to biased predictions due to uneven fold representation.

This study compares protein language models (pLMs) with simple baselines in combinatorial protein design, finding no consistent advantage for pLMs in predicting or generating superior variants.

This study shows that the Gcn1-Gcn20 complex inhibits RQC initiation during transient ribosome collisions, preventing unnecessary protein degradation.

This study shows that nanoparticle vaccines with oligomannose glycans boost B cell responses and protection against malaria sporozoites compared to other glycan types.

The paper develops BERT-T6, a model using transfer learning on ProtBert embeddings to predict T6SEs with high accuracy and performance metrics.