#32 · Genomics

Predicted structures curated into reusable resources for proteome-scale annotation

The common thread is data curation and reuse: structure-derived and literature-derived information is being standardized into resources for proteome-scale use. OpenAntigens provides a free, no-login database of construct-design reports for 5328 human cell-surface and secreted proteins, integrating UniProt, AlphaFold, PDB, and related data. GolgiFold presents a structure-based PPI map of human Golgi complexes that integrates AlphaFold predictions with crosslinking and blue-native proteomics. ProtoPure uses an LLM-based workflow to extract 12 standardized purification parameters from literature across UniProt proteomes. What is new is the scale and standardization of structure-derived data for reuse rather than a new computational method. The leech salivary EV analysis applies ESMFold and Foldseek/TM-align searches to assign a Ly-6/uPAR three-finger fold and flags an unresolved structurally dark fraction, but it is an annotation study rather than a resource. The items do not share a single computational method.