Everyone agrees that data that form the basis of a scientific paper need to be available to readers at the time of publication. But just how raw should such data be? And how much should be released ahead of publication to peer reviewers? These questions can get troublesome when releasing certain data can allow competitors an easy route to results for which the originating researchers have sweated blood. And nowhere is this dilemma more acute than in the data underlying the structures of biological macromolecules and of the complexes that they form.

Structure-factor files are not required upon the submission of a manuscript, but editors may request them to aid the review process.

To labs that solve the structures of proteins and other biological molecules, ‘structure factor’ files are like the reagents of other fields — material that can be used to enable a variety of experiments. Structure factors are the raw data from which atomic coordinates are derived. As such, they are a key aid for reviewers and readers in verifying a structure.

For many years, Nature and its sibling research journals have required that crystallographers deposit the atomic coordinates of their structures in public data banks at the time of publication. In 2000 the International Union of Crystallography weighed the issues and decided that both coordinates and structure-factor files should be released upon publication, and this is now standard practice.

There is less of a consensus over how to balance the interests of authors and the needs of referees in the peer-review process: should we also require structure-factor files upon submission? After gathering the opinions and insights of a slice of the structure community, Nature and its sibling journals have adopted the following policy.

To help authors maintain control of their data before publication, structure-factor files are not required upon submission of a manuscript. However, editors may request them to aid in the review process. Referees who find that evaluation is dependent on factor files should contact the editor, who will obtain the necessary data set.

We believe that this policy balances appropriate control of data access before publication and the need for rigorous review.