AlphaFold Database v6 adds tens of millions of predicted structures for the global life-sciences commons

[Founding Archive] AI20 editorial coverage based on EMBL-EBI / PDBe (2025-10-21). On 21 October 2025, EMBL-EBI’s PDBe announced AlphaFold Database v6, aligning the resource with UniProt release 2025_03. The release adds over 26 million new predicted protein structures and brings the total to more than 241 million (AlphaFold DB v6: 241,070,489 predicted structures, including 40,054 isoforms, versus v4’s 214,683,829). Release notes detail upstream UniProt-driven adds, removals and sequence changes, and clarify that many “unchanged” entries received metadata updates while retaining prior model coordinates. The notice also highlights access changes (preferred entry URLs), API evolution with a dual-support period and a 25 June 2026 sunset for old prediction API fields, and availability of multiple sequence alignments on entry pages. The update is a significant expansion of openly available structural predictions used across biomedical and biotech research workflows worldwide. Why it matters: Open scientific infrastructure quietly determines who can participate in drug discovery, bioengineering and academic research. Database scale, API stability and provenance of updates are operational concerns for R&D leaders as much as model novelty headlines. AI20 perspective: Shared knowledge graphs and open research commons are alliance infrastructure. Expanding globally accessible scientific AI resources strengthens collaborative discovery while reducing dependence on closed, fragmenting data silos. (Category: Healthcare; Region: Global; Tags: AlphaFold; EMBL-EBI; structural biology; open science; research infrastructure. Original source: https://www.ebi.ac.uk/pdbe/news/alphafold-database-release-notes)