Helen M. Berman, director of the RCSB (Research Collabatory for Structural Bioinformatics) Protein Data Bank, and co-authors wrote a Correspondence to Nature ( 463, 425; 2010) to clarify the PDB’s correction procedures and policies in the light of a current investigation. Their letter is reproduced here.
Your News story ‘Fraud rocks protein community’ (Nature 462, 970; 2009) discusses allegations that 12 Protein Data Bank (PDB) entries are based on fabricated data. Pending verdicts on these entries from the US Department of Health and Human Services Office for Research Integrity, we wish to clarify PDB policies and actions.
The PDB archive, which is managed by the Worldwide PDB (wwPDB), houses more than 62,000 entries for macromolecular structure models and their experimental data. It is maintained for the public good. Deposited structures are validated using community-developed standards, and any related corrections are made by depositors before release and publication.
Entries can be replaced on written request from the depositor(s) if better data have become available or the interpretation of existing data has changed. Entries can be withdrawn (that is, rendered obsolete) by the senior author, or by journal editors when the published paper describing the entry is retracted.
An author’s employer (in this case, the University of Alabama at Birmingham) may request removal of an entry, but this request must be fully documented and the original paper describing the entry must be retracted. This ensures due process for the author(s) and the scientific integrity of the PDB archive. To date, the paper describing one of the 12 PDB entries in question has been retracted (J. Biol. Chem. 284, 34468; 2009), and the corresponding PDB entry (PDB code 1BEF) has been made obsolete by the wwPDB at the request of the publisher.
To ensure that PDB entries are validated using state-of-the-art methods, wwPDB validation task forces have been convened for X-ray crystallography and nuclear magnetic resonance spectroscopy. Their recommendations will be reviewed and incorporated into wwPDB’s deposition and annotation procedures.
wwPDB encourages all journals publishing macromolecular structures to stipulate accompanying submission of wwPDB validation reports. These will help editors and referees to assess the reliability of structural data and their interpretation. A few journals have already indicated their interest.
With the support of the structural-biology community, the mission of the wwPDB is to safeguard the integrity and improve the quality of the PDB archive. It is the public availability of atomic coordinates and experimental data that enables errors and possible fabrications to be detected in the first place. Current validation procedures were designed to identify occasional honest mistakes, not to guard against rare cases of malfeasance.
Full authorship of this Correspondence:
Helen M. Berman, Director, RCSB PDB, Rutgers University, Piscataway, New Jersey 08854-8087, USA
Gerard J. Kleywegt, Head, PDBe, EMBL-EBI Hinxton, Cambridge CB10 1SD, UK
Haruki Nakamura, Head, PDBj, Osaka University, 3-2 Yamadaoka, Suita, Osaka 565-0871, Japan
John L. Markely, Head, BioMagResBank, University of Wisconsin-Madison, Madison, Wisconsin 53706-1544, USA
Stephen K. Burley, Chair, wwPDB Advisory Committee, Eli Lilly and Company, San Diego, California 92121, USA
Note from Nature: Nature would like to make clear that the misconduct investigation by the University of Alabama concluded in December 2009 that H. M. K. Murthy acted alone in generating allegedly falsified protein structures. He denies the allegations.