Target hit rate

Figure 12.3 Target Hit Rates (THR) for marketed drugs and Novartis compounds. A TH R of >20% is considered to be promiscuous (black), 5-20% medium promiscuous (hatched) and 0-5% selective (white). 65% of the Marketed Drug set of 293 compounds, including antipsychotics and withdrawn drugs, were selective, whereas only 13% were promiscuous.

Compute the target hit rate (THR) in order to assess each compound for its selectivity or promiscuity across the whole panel of target assays. THR is defined as the ratio of the number of targets hit by a compound (>50% inhibition at a given concentration) to the number of targets tested at that concentration. 

Library designed, on the basis of preexisting information, to generate enhanced activity or hit rate against a particular biological target or target class. 

Lastly, every research project presents a unique combination of parameters (target, assay methodology, hit rate, throughput limit, and so forth) and project team requirements for interesting actives (physiochemical properties, novelty, and so forth). This suggests that an optimal selection scheme should be easily tailored to meet these unique needs, which is certainly not true of a simple Top X approach. 

Even in projects with lower hit rates, carefully constructed filters should be applied, as was shown in the example of the protein-protein interaction target. Given the huge investment in HTS technology by pharmaceutical and biotechnology companies and the importance of the selection of actives to the drug discovery process, the scheme outlined here offers a practicable alternative to the Top X method with tangible and substantial benefits. 

Table 13.1 Top section The mean value of descriptors. Model PC uses physicochemical molecular weight, calculated LogP and target hit based descriptors. Bottom section Sensitivity and rate for compounds originating from different specificity of promiscuous models applied to the disease areas. Middle section Sensitivity and drugs set. Model FP uses fingerprint based specificity of promiscuous models applied to descriptors. Model PC uses physicochemical the test set. Model FP uses fingerprint based based descriptors.

---