ROCKVILLE, Md.—Adding another layer to the explosion of biomolecular databases, Dr. Mike Gilson and colleagues at the University of Maryland Biotechnology Institute (UMBI) developed the Binding Database (BindingDB), a public, web-accessible repository of binding affinities between small molecules and protein drug targets. At present, the repository is home to approximately 14,000 measurements of about 8,000 compounds and 60 targets.
According to Gilson, who is a fellow and professor of UMBI's Center for Advanced Research in Biotechnology, the idea is that a publicly accessible database like BindingDB should speed development of new therapeutics and facilitate development of generally applicable methods of computer-aided drug discovery by academic labs and smaller companies that do not have the in-house resources of larger pharmaceutical companies.
"BindingDB makes it easier to rapidly identify and view compounds and compound classes suitable for a given target," Gilson says. "As the dataset grows more comprehensive, it can be mined for large-scale patterns, such as correlation between affinity and a key chemical descriptor; or for paradoxes and anomalies such as very similar compounds with very different affinities for the same target."
The UMBI group is currently looking to integrate its efforts with related databases, such as the NIH's PubChem repository of chemical information and the IMEx consortium's collection of biomolecular interaction repositories. "My expectation is that ultimately, while each member will focus on extraction and curation of a particular type of data (e.g., protein-protein binding, protein-ligand binding, etc.), we will share the data we extract in order to avoid overlap of effort (and) make each database as complete as possible," Gilson adds.
The group hopes to expand BindingDB by 10,000 records each year and the current data curation activity is on target to maintain this pace. With any effort of this scale, however, public acceptance and participation will be key to success and to that end, the UMBI researchers have established mechanisms by which independent researchers can submit their own experimental data online.
According to Gilson, it would also make his group's curation efforts a lot easier if journal publishers would ask authors to include an electronic representation of their compounds and data along with their papers. "This would not be much more work for the authors, since they probably already have an electronic representation of every compound and indeed a table linking the compounds with the binding data," he says. He is not, however, expecting this to happen any time soon. "I am not optimistic that the ACS [American Chemical Society], which publishes some of the central journals in this area, will adopt such a policy."