MobiDB in 2027: machine-actionable access to intrinsic protein disorder knowledge

MobiDB (https://mobidb.org) integrates structural and functional annotations for intrinsically disordered proteins and regions across UniProtKB. The 2027 release strengthens semantic consistency, programmatic access, and reproducible reuse. A refined controlled vocabulary groups features in a hierarchical way and distinguishes evidence classes and types annotation sources as providers, methods, or aggregation rules. A new versioned REST API, along with a developed MCP server, enables scripts, workflows, and tool-using artificial intelligence systems to retrieve and combine evidence-bearing records through documented operations. The redesigned website applies the same vocabulary to protein entries, documentation, and cross-proteome statistics. Additional developments include standardized Nextflow production workflows, updated RING Protein Data Bank interaction processing, integration of Motif Map of the Proteome and Linear Motif mediated Protein-Protein Interaction Database annotations, and a release-aware catalogue of static datasets for large-scale analysis. Following the recent increase in reference proteomes in UniProtKB, MobiDB now provides a more representative estimate of long-disorder with a median residue coverage of 12.3% in eukaryotes, 1.9% in archaea, 1.7% in bacteria, and 3.4% in viruses; and protein prevalence of 48.7%, 11.9%, 11.2%, and 25.0%, respectively. Together, these developments make disorder knowledge easier to discover, relate to function, and reuse with explicit provenance.