Fortunately, in the last decade the rules of the game have shifted considerably, with certain types of compound bioactivity data entering the public domain: PubChem, probably the best known and one of the largest efforts of this kind so far, is a public repository of predominantly US National Institutes of Health Molecular Libraries Roadmap screening data, ChemBank publicizes screening results from the Broad Institute of Harvard and MIT and DrugBank compiles more narrow but also much more in-depth information on a set of thousands of US Food and Drug Administration–approved drugs, and so on. However, 18 January, with the official release of the ChEMBL database (ChEMBLdb, version Chembl_02, http://www.ebi.ac.uk/chembldb/index.php) at the European Bioinformatics Institute, marked an important addition to the public domain: a database containing large-scale, manually curated (and thus usually higher quality) data covering a large amount of chemical space with largely standardized data taken from dose-response experiments and with annotated literature references.
Funded by a Strategic Award from the Wellcome Trust for the next 5 years, the ChEMBLdb and related tools are being developed by the ChEMBL group, which is headed by John Overington at the European Bioinformatics Institute. The initial database, originally called StARlite, was acquired from Galapagos in 2008. The current release of the database contains about 2.4 million activities of 622,824 compounds, measured against 7,192 targets, 4,364 of which are protein targets (the others are cell lines or organisms). This release of ChEMBLdb also includes 24,000 natural products, which is particularly noteworthy because databases linking natural products to protein targets have been notoriously difficult to find in the past. Data are abstracted from a total of nearly 34,000 publications, taken from 12 prominent medicinal chemistry journals such as the Journal of Medicinal Chemistry. Hence, coverage of public bioactivity data, while not complete, is by far the most comprehensive ever seen in a public database.
This is a preview of subscription content, access via your institution
