Biomolecular Object Network Databank: Difference between revisions

Revision as of 18:36, 28 February 2007

Revision as of 18:37, 28 February 2007

Line 4:

==Biomolecular Object Network Databank (BOND)==

The Biomolecular Object Network Databank (BOND), which integrates the original blueprint initiative databases as well as other databases, such as Genbank, combined with many tools required to analyze these data. Annotations links for sequences, including taxon identifiers, redundant sequences, Genome Ontology descriptions, Online Mendelian Inheritance in Man identifiers, conserved domains, data base cross-references, LocusLink Identifiers and complete genomes are also available. Bond facilitates cross-database queries and is an open access resource which integrates interaction and sequence data[2].

===Small Molecule Interaction Database (SMID)===

The Small Molecule Interaction Database is a database containing protein ___domain-small molecule interactions. It uses a ___domain-based approach to identify ___domain families, found in the Conserved Domain Database (CCD), which interact with a query small molecule. The CCD from NCBI amalgamates data from several different sources; Protein FAMilies (PFAM), Simple Modular Architecture Research Tool (SMART), Cluster of Orthologous Genes (COGs), and NCBI’s own curated sequences. The data in SMID is derived from the Protein Data Bank (PDB), a database of known protein crystal structures.

Line 17:

SMID-BLAST is a tool developed to annotate known small-molecule binding sites as well as to predict binding sites in proteins whose crystal structures have not yet been determined. The prediction is based on extrapolation of known interactions, found in the PDB, to interactions between an uncrystallized protein with a small molecule of interest. SMID-BLAST was validated against a test set of known small molecule interactions from the PDB. It was shown to be an accurate predictor of protein-small molecule; 60% of predicted interactions identically matched the PDB annotated binding site, and of these 73% had greater than 80% of the binding residues of the protein correctly identified. Hogue, C et al. estimated that 45% of predictions that were not observed in the PDB data do in fact represent true positives[3].

===Biomolecular Interaction Network Database (BIND)===