Data quality in the PDB archive
John Westbrook, K. Henrick, Zukang Feng, Dimitris Dimitropoulos, Shuchismita Dutta, C.L. Lawson, John L. Markley, H. Nakamura, J. Swaminathan, Eldon L. Ulrich, Jasmine Y. Young, Helen Miriam Berman · Acta Crystallographica Section A Foundations of Crystallography · 2008
The quality of the carbohydrates in PDB entries is rather poor compared with the protein parts [1], with about 30% of the PDB entries with carbohydrates being erroneous [2].The main reasons for this are the complexity of carbohydrates and the lack of check programs for carbohydrate 3D structures.Recently, such check tools were established.Pdb-care (PDB CArbohydrate REsidue check) [www.glycosciences.de/tools/pdb-care/]checks if the carbohydrate residue names used in a PDB file match the monosaccharide units present in the structure [3].Furthermore, the connectivities given in a PDB file are checked.These bond checks can be applied not only to carbohydrates but to any residue.To evaluate the conformation of a carbohydrate chain, plots of the phi / psi angles of the glycosidic linkages can be used, similar to the Ramachandran Plot for proteins [4].However, the preferred glycan torsions depend on the involved residues and the linkage position.Thus, separate plots have to be generated for each type of disaccharide fragment in the structure.This can be done with carp [www.glycosciences.de/tools/carp/].Detected torsions can be compared with either all torsions of the respective type that are present in the PDB or with computed energy maps taken from GlycoMapsDB [www.glycosciences.de/modeling/glycomapsdb/] [5].Use of these tools will help to increase the reliability of carbohydrate 3D structures.