In recent years, awareness of the need for effective management, deposition and validation of research data has been growing amongst national science policy makers, funding bodies, and publishers, database managers and librarians. For example, the UK Research Data Service published in 2009 a report "The data imperative: Managing the UK's research data for future use", assessing the status of government-led initiatives to manage digital research data as a key component of national e-infrastructure, and stressing the need for more and better sharing of research data. The report of the e-IRG Data Management Task Force provided a broad European perspective on these issues and emphasized the need for quality assurance in data management. The September 2009 Nature editorial, "Data's shameful neglect", reinforced the need for integrating publications and supporting data as a full record of the outcome of a research project.
Crystallography has long led the field in this respect. The routine sharing of structure models, and the structure-factor data from which they were derived, has been second nature to small-molecule crystallographers since the birth of the subject. In more recent decades, the growing field of biological macromolecular crystallography has adopted policies that balance the desire for privileged use of a hard-won data set against the benefits to the wider community of its early dissemination. With high-quality curated databases such as the Cambridge Structural Database and the Protein Data Bank, central archives exist with comprehensive holdings of carefully validated data. The journals of the IUCr have routinely made freely available structure models and structure factors, in the form of standard machine-readable data files since the mid-1990s.
However, more can yet be done. In the area of small-unit-cell crystallography (inorganic structures, organic and organometallic small molecules), not all journals that publish crystal structures yet require authors to deposit their structure factors; and not all structural models are technically reviewed for consistency and chemical reasonableness, using for example the validation routines employed in the IUCr checkCIF service.
In an effort to encourage best practice in data validation and deposition, the IUCr President and the Editor-in-Chief of the IUCr journals have sent a letter to publishers and editors of journals reporting such crystal structure determinations. This letter invites wholehearted participation in a programme of encouraging authors to provide machine-readable CIF and structure-factor files for every submitted structure, and to make full use of services such as checkCIF to avoid errors and ensure the quality of the reported science.
The text of this letter is reproduced below.
Publication standards for crystal structures
On behalf of the International Union of Crystallography (IUCr), we are writing to publishers and editors of scientific works reporting crystal structure determinations about the importance of retaining and validating supporting experimental data sets.
The IUCr has for many years contributed to the advancement of crystallography in all its aspects, disseminating scientific achievements and standards in the field, and providing high-quality publishing and data services. At present, eight journals (and a number of other publications) are published on a regular basis to serve the scientific community at a high professional level.
Crystallography has recently seen a number of high-profile retractions of papers from the peer-reviewed literature, both in the small molecule and in the macromolecular areas. These cases frequently involved actions of scientific misconduct such as manual alterations to cell constants and atom types to produce what appeared to be genuine structure determinations of new compounds. Such cases are not always easy to detect, but once published, they compromise the quality and dependability of the scientific literature.
The response of the IUCr has been to develop and implement measures that will help to ensure that such cases are identified before publication. It has become clear in the process that many of these problems could not have been readily discovered without the availability of the structure-factor files. Deposition of such files is currently mandatory for publications in the macromolecular field, but only required by a small number of journals publishing inorganic and small molecule structures. We therefore strongly recommend that all journals publishing crystal structures require authors to provide structure factors for checking and archiving.
In order to help editors and publishers follow this recommendation, the IUCr checkCIF service at http://checkcif.iucr.org has been modified to allow checking of crystallographic information files (CIFs) and structure factors. In addition, the Cambridge Crystallographic Data Centre will also now allow authors to deposit structure-factor files.
In practical terms, the important points are as follows:
(1) Encourage authors to provide a CIF and structure factors in machine-readable format for each submitted structure. These might be collected as supplementary material or deposited with the Cambridge Structural Database.
(2) Ask the authors to check the CIF and structure factors before submission using the checkCIF service.
(3) Use the checkCIF service to create a validation report for each submitted structure and make this report available to the referees.
(4) Ensure that referees and editors know how to interpret checkCIF reports correctly.
The IUCr is keen to help you with the implementation of these recommendations. We are happy to provide advice on the collection and checking of structure factors, and you are invited to contact Peter Strickland (email@example.com) at the IUCr Office to discuss this matter further. We are also considering organising a short meeting addressing these issues at our Congress in Madrid in August 2011 (http://www.iucr2011madrid.es/) and would be keen to invite you and your editors to such a meeting. In the longer term, we will also look into how efforts in teaching (summer schools, teaching sessions and materials etc.) might be used in this context.
Sine Larsen, President of the IUCr
Gernot Kostorz, Editor-in-chief, IUCr Journals
Comments and suggestions are welcomed from the wider community for further initiatives to make crystallographic data an exemplar of the best practice that is increasingly demanded by research funders, science policy bodies, and the public at large.