protein sequence

The paper introduces a domain-centric method using physico-chemical properties to classify proteins, improving accuracy over traditional phylogeny-based approaches.

This paper introduces a modular genetic code expansion approach to install diverse lysine acylations on proteins, enabling systematic studies of their functional impacts.

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.

The paper evaluates AlphaFold2's representation for improving sequence-based prediction tasks, showing superior performance over traditional features.

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

EpicTope predicts and validates epitope tagging sites by analyzing structural and functional features, showing effectiveness in zebrafish proteins.

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

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

This paper develops a computational method using graph neural networks and Metropolis Monte Carlo sampling to design high-affinity nanobodies, achieving a Pearson correlation of 0.62 on benchmark.

PMScanR identifies and analyzes protein motifs, evaluates substitution impacts, and provides visualization tools for motif analysis in large datasets.

This paper introduces a deep learning model that predicts if two proteins catalyze the same reaction using ESM-C latent representations and attention mechanisms, outperforming existing methods.

This paper introduces MultiAMP, a framework for predicting and designing antimicrobial peptides by integrating sequence and structural information, achieving state-of-the-art performance and discovering novel peptides.

This paper introduces DETANGO, a deep learning framework that disentangles mutation effects on protein stability and function, identifying stable-but-inactive variants and functionally critical residues.

CARTiBASE curates and standardizes over 10,000 CAR sequences into a single resource, enabling interactive browsing, similarity search, and analysis of CAR design trends.

This paper introduces CALM, a model that predicts antibody-antigen binding from sequences using contrastive learning, achieving 7% top-1 retrieval on a test set.

The paper identifies conserved and variable protein complex components across 31 eukaryotes using orthology mapping and reveals duplications and reductions in specific complexes.

The paper identifies 1296 human genes with dual-coding regions (DCRs) and shows their conservation in mice, tissue-specific expression, and functional implications in gene regulation.

The paper develops a sequence-based method to identify structure-disrupting mutations using protein language models, reducing the need for expensive structure prediction.

This study fine-tunes a protein language model with generated CAR sequences to predict CAR-T cell activity, showing improved performance with optimized training parameters.

The paper designs CsgA variants with altered β-strand lengths to explore structural plasticity in microbial β-solenoids, revealing how strand length affects nanofiber stability and material properties.

The paper introduces a dataset and a graph-transformer model to predict IDR-protein binding affinity, achieving high accuracy and generalizing to designed binders.

The paper develops a rule-based model to generate synthetic mitochondrial targeting sequences in yeast and demonstrates their functionality in directing proteins to mitochondria.

This paper introduces a framework that integrates biophysical priors into protein embeddings to improve DeltaDeltaG prediction, showing that general-purpose encoders can outperform specialized models.

CrossAffinity predicts protein-protein binding affinity using a sequence-based cross-attention model, outperforming both structure-based and sequence-based methods on multiple test sets.

PATHOS predicts variant pathogenicity by integrating protein language models and biological features, achieving high accuracy on clinical and ClinVar datasets.

This paper develops light-controlled chemomechanical networks using Tcb2 protein to achieve rapid, repeatable actuation and dynamic self-assembly.

The paper introduces a sparse learning method to identify phylogenetically informative sites in alignments, showing it matches full alignment results and provides an objective metric for evaluating alignment filtering tools.

PEINT simulates realistic protein evolution by modeling sequence changes over time, generating functional proteins that retain enzymatic activity.

The paper annotates 200 insect genomes using BRAKER3 and VARUS to enable consistent comparative genomics analyses across species.

FeatureDCA improves protein sequence generation by conditioning DCA on principal components, achieving accurate structural and functional results across multiple families.