A Hit Generation by Virtual Screening

In conventional approaches, databases of enimierated molecules are searched. External vendor databases consist of 10—20 million compoimds, which are usually physically available [9]. VS and purchasing the virtual hits from the external vendors dramatically expands the in-house screening pools that at maximum contain only a few miUion compounds. Here at Boehringer Ingelheim on a routine basis, we search in external vendor databases. Thus, we are complementing 

By means of FTress similarity descriptors, extremely large chemical spaces (on the order of 10 °) are accessible for virtual screening techniques. Besides the FTrees, recently other techniques have also been published [11,12]. 

As mentioned above, structure-based virtual screening is the dominating technique in the field. Lyne et al [64] from AstraZeneca reported a structure-based docking procedure. They performed a virtual screening campaign for checkpoint kinase-1 (Chk-1 kinase) inhibitors. Inhibition of Chk-1 allows cells with damaged DNA to enter lethal mitosis. Therefore, Chk-1 inhibitors are of interest especially in combination with therapies that induce DNA damage. 

Besides the mentioned example [58-63], a large number of published successful virtual screening examples exist It is difficult to compare the published data because the definition of potency thresholds for a hit and the term successful is very arbitrary. Thus, it is very hard to estimate data like success probabilities. From the authors experience, besides the successful cases a significant proportion of VS runs are also not successful in identifying new potent hits. Especially when using VS methods in projects where a low hit rate is expected (e.g., protein-protein interactions), the enrichments are occasionally not high enough. 

Ligand-based VS in Nonenumerated Virtual Chemical Spaces 

---