protein design

This study shows that general-purpose LLMs like DeepSeek V3.2 can generate protein sequences with desired structures and improve solubility without biological training.

This paper engineers an energy-efficient pentose phosphate pathway in yeast to boost fatty acid production without compromising cell growth, achieving a record 41.7 g/L titer.

PIONEER is a yeast-based platform for screening genetically encoded protein regulators by displaying them in the periplasmic space to assess their function through signaling.

This study introduces CatOpt, a deep learning model that predicts enzyme optimal pH and designs point mutations to improve acid resistance, achieving a 7% activity increase in a specific enzyme.

An all-atom diffusion model unifying biomolecular structure prediction and sequence-structure generation designs experimentally validated nanobodies, mini-proteins, and linear or cyclic peptides across 26 diverse targets.

This study examines how protein tags affect condensate formation in different contexts and identifies charge as a key factor influencing condensation behavior.

cpiVAE improves cross-platform proteomics imputation between Olink and SomaScan, achieving higher correlation than KNN and WNN, and enabling better phenotype associations.

The paper introduces a framework using MSA-Transformer to generate hybrid proteins between specified source and target proteins, producing sequences with higher consistency and plausibility than random baselines.

STAMPS designs novel mammalian signal peptides to enhance protein secretion, outperforming existing benchmarks in multiple cell lines.

CIRPIN identifies protein homologs related by circular permutation using a permutation-invariant graph neural network, uncovering thousands of novel pairs and revealing natural PDZ domain permutations.

This study identifies the differential roles of SRSF1's RNA recognition motifs in protein interactions using SILAC proteomics.

Genolator is a multimodal large language model that combines genomic and structural data with natural language to interpret protein function, outperforming existing models in accuracy.

The paper develops a single-component optogenetic system using DmCRY to control gene expression and generate human portrait images from cultured cells.

SignalGen predicts optimal signal peptides for protein expression by integrating mature protein sequences, organism, and localization using a Latent Residual Transformer.

The paper investigates TDP-43 SUMOylation and its role in DNA damage repair under oxidative stress.

HalluDesign optimizes and designs proteins using iterative structure hallucination and sequence design, achieving high success rates and structural accuracy in various tasks.

This paper introduces GPF, a lightweight and interpretable method for predicting protein functional properties from DNA sequences, validated on GFP mutations.

SpecLig generates target-specific ligands with high affinity and specificity by integrating hierarchical modeling and energy guidance, improving therapeutic potential and reducing off-target risks.

The paper introduces a system for predicting tissue-specific E3 ligases for targeted protein degradation using a BERT-based model and integrated biological data.

This paper designs antibodies de novo with atomic precision using RFdiffusion and yeast display, validated by cryo-EM and biophysical methods.

An atom-level graph neural network that denoises corrupted structures generates small-molecule and protein-ligand conformational ensembles, and its active-site pre-organization metric enriched for more active designed retroaldolases.

MotifAE discovers functional sequence patterns from protein language models through an unsupervised framework, improving motif identification and enabling stability prediction.

This paper designs ultrastable helical proteins using AI and chemistry, achieving high mechanical, thermal, and chemical stability.

This study clones fluorescent proteins with an antimicrobial peptide to enable real-time visualization of its mechanism of action.

This paper introduces NUWA, an mRNA language model that enables unified sequence perception and generation across diverse species and tasks.

PLiCat uses large language models to classify protein-lipid interactions across eight lipid categories and predict lipid-binding sites and mutation impacts.

The paper introduces MIRAGE, a multi-modal generative model that integrates protein sequence, interaction, and localization data to improve function prediction and module detection.

The paper introduces FcBAR, a method to detect protein-protein interactions in situ for recombinant proteins, and identifies factors that enhance rituximab production.

This paper describes a high-throughput protocol for expressing, exporting, and assaying recombinant proteins from E. coli in a single multiwell plate, enabling direct use in enzymatic assays without purification.

The paper identifies DDX3X as a dual switch regulator of mRNA translation, promoting or suppressing translation in non-stressed cells and switching roles in acute ER stress.