Overview of SAR Knowledge Bases

As indicated above, the focus of this review is on curated databases containing chemical structure and biological activity taken from literature sources. We nevertheless recognise the existence of a growing number of databases that also contain such structure and activity data but which do not fall within our primary scope. Of particular relevance here are databases that contain raw screening data, often generated by publicly-funded research initiatives. Two prominent examples are Pubchem [4] and Chembank [5]. 

2 million chemical structures with data from more than 2500 high-throughput assays and approximately 90 small-molecule microarray assays. Detailed information is provided, including plate locations ChemBank also provides a number of tools for the visualisation and analysis of this data. 

Both Pubchem and ChemBank are publicly available over the Internet as such they can be considered successors to the databases hosted at the National Cancer Institute (NCI). The NCI data set contains more than 250,000 molecules together with biological activity obtained from their anti-cancer and HIV screens. For many years this data set was the only one of any significant size available publicly and as such it has been widely used to test new data-mining approaches [8]. 

Advances in structural biology have resulted in an exponential increase in the number of proteins and protein-ligand complexes for which detailed 3-dimensional information is available, typically from X-ray crystallography or NMR. This has been accompanied by a continual desire to use structural information in drug design. Three general applications can be identified [12]  

4 Other databases with binding or affinity data 

---