Changing the Substrate Specificity of an Enzyme

For lead finding, the use of chemical compound collections and combinatorial libraries is indispensable to every pharmaceutical company for a timely and efficient discovery of novel bioactive entities [5]. Until recently, most attention was paid only to internal historic compound collections, maintained over several years of medicinal research [6]. Flowev-er, those libraries typically contain a limited number of structural classes as the result of past research projects which focused on biological activity islands .Today there is a growing interest in techniques that expand structural and - even more important - biological diversity, this perhaps being accomplished by the direct acquisition of an external compound collection, or by the use of internal or external synthesis capacities [7]. 

The evaluation of chemical databases and the rational design of combinatorial libraries and subsets is important to optimize resources for a successful discovery strategy. Although today s synthesis and assay technology allows the processing of large libraries, their production and testing must be carried out in an efficient and cost-effective way 

If additional information is available, such as 3D-QSAR models [17], pharmacophore hypotheses [18], or 3D target protein structures [19], the library design should be focused towards those motifs or to regions shown to be essential in explaining the fundamental 

All those design alternatives require a virtual library as a fundamental basis, which refers to an electronic description of all possible compounds, which might be synthesized by a short sequence of specific transformations with particular reagents and one or multiple scaffolds. Appropriate fast and relevant searching techniques are available today, which can be used for selection and design of actual synthesis subsets in accordance with the level of biological knowledge [27,28] from such a parent library. 

Subsequently, for every validated primary screening hit from rationally designed libraries, the nearest neighbors must be identified using similarity searching [29, 30] based on the same descriptors as were used for design. If not already available, this subset of nearest neighbors must be synthesized and tested in a second iteration to establish the valid- 

---