PubChem (57) is the informatics backbone for the NIH Roadmap Initiative on molecular libraries and focuses on the chemical, structural and biological properties of small molecules, in particular their roles as diagnostic and therapeutic agents. A suite of three Entrez databases, PCSubstance, PCCompound and PCBioAssay, contain the substance information, compound structures and bioactivity data of the PubChem project. The databases hold records for 61 million substances containing 25 million unique structures. More than 920 000 of these substances have bioactivity data in at least one of the 1700 PubChem BioAssays. The PubChem databases link not only to other Entrez databases such as PubMed and PubMed Central but also to Entrez Structure and Protein to provide a bridge between the macromolecules of genomics and the small organic molecules of cellular metabolism. The PubChem databases are searchable using text queries as well as structural queries based on chemical SMILES, formulas or chemical structures provided in a variety of formats. An online structure-drawing tool (pubchem.ncbi.nlm.nih.gov/search/search.cgi) provides a simple way to construct a structure-based search.
