protein sequence

AI-IDP predicts conformational ensembles of disordered proteins by combining deep learning with physical restraints, revealing common structural features across proteomes.

This study analyzes PRPS enzyme evolution across domains of life, identifying new classes and regulatory mechanisms through phylogenetic and biochemical approaches.

The paper resolves a long-standing paradox in lysozyme evolution by showing that recombination and incomplete lineage sorting explain apparent molecular convergence without requiring fixation of deleterious intermediates.

This study shows that an A2062U mutation in 23S rRNA improves ribosome activity for translating sequences with challenging monomers like proline and β2-hydroxy acids.

The paper introduces PE-TFD, a method using protein embeddings and machine learning to discover novel protein functions, demonstrating its effectiveness in identifying hydrogenases not found by traditional methods.

ProtAttn-QuadNet predicts protein-protein interactions using ProtBERT embeddings, achieving high accuracy on benchmark datasets and outperforming existing methods.

This paper designs intrinsically disordered proteins using a physics-guided machine learning framework to encode complex signal processing and condensate properties.

The paper quantifies and predicts multiple sequence alignment difficulty using AlDiScore, achieving high R2 scores on unseen datasets.

This study examines the expression patterns of Slug and Snail in breast cancer subtypes, linking their expression to proliferation status and clinical outcomes.

ProSiteHunter improves protein binding site prediction by integrating sequence-based features and a fine-tuned language model, achieving better performance than existing methods.

This paper evaluates two mechanistic interpretability methods on a protein language model to identify features distinguishing Class A beta-lactamase from other classes.

The paper investigates IDR-IDR interactions in synthetic systems to assess their ability to mediate specific condensate formation, finding high promiscuity among some IDRs and specific recognition in others.

This paper identifies a novel role for the R2T complex in early box H/ACA snoRNP maturation by characterizing interactions between SHQ1, RuvBL1:RuvBL2, and other factors.

This paper develops a hierarchical sequence-based framework to infer distributed protein organization, identifies novel ciliary proteins, and enables synthetic protein design.

This paper introduces STAMP, a machine learning framework for predicting antimicrobial peptide efficacy across bacterial species, achieving strong predictive performance and real-world validation.

The paper identifies OLFM4 as a marker of high-risk gastric intestinal metaplasia, showing its restricted localization to epithelial glands and focal expansion with disease severity.

The paper introduces i-POPFLEX, a modular cell-free protein synthesis system that improves protein yield and reduces costs compared to commercial kits.

PEPE efficiently extracts multimodal protein embeddings from language models, enabling scalable and memory-efficient generation of large embedding datasets for downstream biological tasks.

The paper develops ML tools to improve genotype-phenotype prediction for opsin genes by introducing new methods and software.

This paper introduces FunBind, a multimodal model for protein function prediction that improves accuracy and enables zero-shot prediction of novel functions.

The paper develops a framework to discover new protein paralogs by integrating sequence, structure, and embedding-based methods with family-specific models, identifying 7 new protease and 3 new kinase paralogs.

This study evaluates AAV transgene cassettes for expressing 10-1074 antibody in nonhuman primates and mice, showing promoter and WPRE effects on expression levels and immune responses.

The paper improves viral protein clustering by combining diversified sequence profiles with structural information, revealing a more accurate picture of viral protein diversity.

SSPSPredictor predicts phase-separating proteins using sequence and structure data, showing higher LLPS propensity in intrinsically disordered proteins and pathogenic variants.

This paper introduces ProtEGNN, a model combining protein language models with geometric equivariance to improve protein property and function prediction with fewer parameters than fine-tuned PLMs.

This paper introduces OCDesign, a curriculum-guided framework for multi-property protein design, showing that staged objective introduction improves success rates compared to one-shot optimization.

LucaPhylo infers phylogenetic trees from unaligned protein sequences using hyperbolic geometry and protein language models, achieving accuracy comparable to alignment-based methods for divergent RNA viruses.

This study compares RAM-NJ and SAM-NJ phylogenies using protein foundation models to reconstruct species relationships without conventional substitution models or multiple sequence alignments.

The paper uses a mechanistic model to explain the bell-shaped protein production profile in the PURE system, showing that excess EF-Tu sequesters tRNAfMet, and confirms this by experimental validation.

Deep-Palm is a deep learning framework that accurately predicts S-palmitoylation sites by integrating sequence, physicochemical, embedding, and structural features, outperforming existing methods.