PEDANT genome database: 10 years online

Abstract

The PEDANT genome database provides exhaustive annotation of 468 genomes by a broad set of bioinformatics algorithms. We describe recent developments of the PEDANT Web server. The all-new Graphical User Interface (GUI) implemented in Javatrade mark allows for more efficient navigation of the genome data, extended search capabilities, user customization and export facilities. The DNA and Protein viewers have been made highly dynamic and customizable. We also provide Web Services to access the entire body of PEDANT data programmatically. Finally, we report on the application of association rule mining for automatic detection of potential annotation errors. PEDANT is freely accessible to academic users at http://pedant.gsf.de.

Figure 1

Navigation tree and report page for the hypothetical secreted protein of Helicobacter pylori (gi_15645712). This figure also illustrates the application of association rule mining for finding potential annotation errors (see text). A number of domain assignments have been marked as suspicious. The InterPro domain IPR001440 and PFAM domain PF00515 were erroneously assigned to this gene product based on a weak similarity hit to a single TPR_1 repeat whereas the domain definition requires at least three repeat copies. Note that all other features marked as suspicious are in fact annotated correctly; the method does not detect annotation errors as such, but rather incompatible feature combinations which might include annotation errors.