COMe: the ontology of bioinorganic proteins

Many characterised proteins contain metal ions, small organic molecules or modified residues. In contrast, the huge amount of data generated by genome projects consists exclusively of sequences with almost no annotation. One of the goals of the structural genomics initiative is to provide representative three-dimensional (3-D) structures for as many protein/domain folds as possible to allow successful homology modelling. However, important functional features such as metal co-ordination or a type of prosthetic group are not always conserved in homologous proteins. So far, the problem of correct annotation of bioinorganic proteins has been largely ignored by the bioinformatics community and information on bioinorganic centres obtained by methods other than crystallography or NMR is only available in literature databases.

COMe (Co-Ordination of Metals) represents the ontology for bioinorganic and other small molecule centres in complex proteins. COMe consists of three types of entities: 'bioinorganic motif' (BIM), 'molecule' (MOL), and 'complex proteins' (PRX), with each entity being assigned a unique identifier. A BIM consists of at least one centre (metal atom, inorganic cluster, organic molecule) and two or more endogenous and/or exogenous ligands. BIMs are represented as one-dimensional (1-D) strings and 2-D diagrams. A MOL entity represents a 'small molecule' which, when in complex with one or more polypeptides, forms a functional protein. The PRX entities refer to the functional proteins as well as to separate protein domains and subunits. The complex proteins in COMe are subdivided into three categories: (i) metalloproteins, (ii) organic prosthetic group proteins and (iii) modified amino acid proteins. The data are currently stored in both XML format and a relational database and are available at http://www.ebi.ac.uk/come/.

COMe provides the classification of proteins according to their 'bioinorganic' features and thus is orthogonal to other classification schemes, such as those based on sequence similarity, 3-D fold, enzyme activity, or biological process. The hierarchical organisation of the controlled vocabulary allows both for annotation and querying at different levels of granularity.