Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain
the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in
Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles
The rise of AlphaFold may not have put structural biologists out of work, but it has undeniably transformed the field by allowing researchers to nearly instantaneously generate high-quality predictions of structures for many proteins that would take months or years to experimentally determine. But we are still far from a full solution to the ‘protein structure prediction problem’. “There's a ridiculous amount of structural disorder in proteins,” says Kamil Tamiola, CEO and co-founder of London, England–based Peptone. The total number of such proteins remains an open question, but he estimates that as many as 18,000 human proteins—roughly 75% of the proteome—may contain stretches of 70 or more amino acids that lack a stable and defined structural conformation. Such ‘intrinsically disordered’ regions currently remain an insurmountable challenge for AlphaFold or any other structure-prediction program.