structural-biology

Chamaileon introduces a framework for multi-target and multi-state protein binder design using contextualized sequence-structure co-modeling and mixed sampling, achieving adaptability across diverse conformational landscapes and multi-target requirements.

This paper investigates discrete Ricci curvature on protein contact graphs as a lightweight structural descriptor for fold classification, showing that a 22-dimensional curvature feature outperforms mean-pooled ESM-2 embeddings on CATH and SCOPe benchmarks.

Two Nature papers from Baker and Veesler labs describe AI-designed capsids that achieve quasi-symmetry, enabling larger protein shells than previously possible, clarifying that they designed architecture, not actual viruses.

Introduces PairSAE, a method that adapts sparse autoencoders to interpret pairwise representations in protein co-folding models, enabling the discovery of interpretable features that align with biological annotations and predict binding affinities.

A blog post from Ligo discussing the redundancy of natural protein folds and the challenges of scaling structural data for generative biomolecular models, referencing AlphaFold3 and other recent models.

A user reports that GPT 5.5 successfully conducts autonomous research in structural biology, improving AlphaFold2's performance after 150+ hours of goal mode.

This paper proposes SoftBlobGIN, a framework that enhances the interpretability of protein language model representations by projecting them onto contact graphs for structure-aware message passing. It demonstrates improved performance on enzyme classification and binding-site detection while providing auditable structural explanations.

Researchers used AlphaFold and cryo-EM to map the structure of the apoB100 protein, which forms bad cholesterol, marking a significant breakthrough in understanding heart disease.