Nature is raising ire over the lack of code availability for Google DeepMind’s protein folding paper

through Nature

A group of researchers is busy Nature tasked with publishing a paper earlier this month about Google’s protein folding prediction program DeepMind, without the authors having to publish the code behind the work.

Roland Dunbrack, of Fox Chase Cancer Center in Philadelphia, submitted the paper to a peer review but was “not given access to the code during the review,” write the authors of a letter submitted yesterday, May 14, to Nature – including Dunbrack – write “despite repeated requests.”

a Nature podcast said that AlphaFold3 – unlike AlphaFold2 – “can accurately predict protein-molecule complexes containing DNA, RNA and more. Although the new version is limited to non-commercial use, researchers are excited about the expanded range of predictive capabilities and the prospect of faster drug discovery.”

Not everyone was enthusiastic. The authors of the letter, co-author Stephanie A. Wankowicz of the University of California, San Francisco, told Retraction Watch was submitted to Nature today, write that they were “disappointed with the lack of code, or even executables, included with the publication of AlphaFold3. in Nature.” They continue:

Although AlphaFold3 extends the capabilities of AlphaFold2 with small molecules, nucleic acids and chemical modifications, it was released without the means to test and deploy the software in a rapid manner. This is inconsistent with the principles of scientific progress, which depend on the community’s ability to evaluate, use, and build on existing work. The high-profile publication promotes opportunities that remain hidden behind the doors of its parent company.

The authors, circulating the letter for additional signatures, write that “the limited availability of the model on a hosted web server, limited to ten predictions per day, limits the ability of the scientific community to verify the broad claims of the findings or apply the predictions on a large scale. In particular, the inability to make predictions about new organic molecules that resemble chemical probes and drugs, one of the central claims of the paper, makes this method impossible to test or use.”

A news item from May 8 by the independent team of journalists from Nature noted the limitations. Nature editor-in-chief Magdalena Skipper told Retraction Watch:

Nature has long-standing policies designed to facilitate the availability of data, materials, and code upon reasonable request. While we seek to increase transparency at every opportunity, Nature accepts that there may be circumstances where research data or code is not openly available. When making a decision about data and code availability, we consider many different factors, including the potential biosafety implications and the ethical challenges this poses. In such cases, we work with the authors to provide alternatives that support reproducibility, for example by providing pseudocode, which is made available to the reviewers during peer review.

As noted in the code availability statement in the article: AlphaFold 3 is available as a server for non-commercial use only at https://www.alphafoldserver.com, with restrictions on permitted ligands and covalent modifications. Pseudocode describing the algorithms is available in the Supplementary Information.

However, the pseudocode will require “months of effort to be converted into workable code that approaches performance, wasting valuable time and resources,” the letter’s authors write. “Even if such a reimplementation is attempted, limited access raises questions about whether the results can be fully validated.”

The authors of the letter continue:

When journals fail to enforce their written policies regarding making code available to reviewers and alongside publications, they demonstrate how these policies are unfairly enforced and how editorial decisions do not reflect the needs of the scientific community . While there is an ever-changing landscape of how science is conducted and communicated, journals must uphold their role in the community by ensuring that science is reproducible when disseminated, regardless of who the authors are.

Like Retraction Watch? You can have one tax-deductible contribution to support our worksubscribe to our for free digest daily or paid weekly update, follow us on Twitterlike us on Facebookor add us to your RSS reader. If you find a retraction, that is not in the Retraction Watch Databaseyou can Let us know here. For comments or feedback, please send an email to [email protected].

Leave a Reply

Your email address will not be published. Required fields are marked *