The AlphaFold database expands to over 200 million structures

node

EMBL-EBI and DeepMind expanded the database roughly two hundredfold, to predicted structures for more than 200 million protein sequences — effectively the whole of UniProt, covering almost every organism whose genome has been sequenced. A 2024 database paper reports coverage of over 214 million sequences. This is the moment the change becomes structural rather than exemplary: predicted structure stopped being something obtained for a protein of interest and became a property the catalogue already had.

A node is not a place. Drawing it on a map would assert something about the world that no stored fact supports.