Torsten F. Schwede

Stoic predicts protein complex stoichiometry using protein language model embeddings and graph neural networks, achieving fast and accurate results for homomeric and heteromeric complexes.

This study evaluates four protein-ligand cofolding methods using a new benchmark dataset, finding they primarily memorize ligand poses rather than predict them de novo.

This study identifies natural genetic variants in human therapeutic target antigens that can disrupt antibody binding, leading to resistance to antibody-based therapies.

This paper introduces a new 20-letter alphabet, TEA, derived from protein language models, enabling efficient homology searches comparable to structure-based methods without structural data.

AlphaPulldown2 provides a pipeline for high-throughput protein structural modeling with automated workflows and improved data management.

PLINDER provides a large, annotated PLI dataset with diverse annotations and improved evaluation splits to better assess computational methods for drug design.

This study analyzes BKPyV sequence variations in Vp1 and LTag using computational methods, finding high variability in Vp1 that may impact immune control and vaccine design, while LTag shows less variability but potential effects on T-cell epitopes.

This paper introduces an embedding-based alignment method using protein language models to detect structural similarities in the twilight zone, outperforming existing approaches without training.

This study evaluates AlphaFold2's accuracy in predicting outer membrane beta-barrel structures, showing high accuracy even for novel topologies without templates.

ModelCIF extends the PDBx/mmCIF standard to represent predicted 3D protein models, enabling FAIR data sharing for computational structural biology.

3D-Beacons provides a federated network to unify access to protein structure data from multiple sources, enabling comparative analysis of models.

The paper reports the high-resolution cryo-EM structure of urease from Yersinia enterocolitica, revealing the bimetal active site and enabling drug development insights.

This paper describes OpenEBench, a continuous automated benchmarking system for bioinformatics methods, tools, and web services to improve reproducibility and interoperability.