Design for Developability or Drug-likeness

As discussed above, many methods have been developed to optimize many properties simultaneously, including ADMET filters. 

Samiulla et u/. report the design of natural product scaffolds that were then used in a small library of vasicinone derivatives. A key aspect of the design was the incorporation of parameters intended to result in improved ADME properties by including the filters proposed by Lipinski and Veber et BCUTs were used to optimize the diversity of the scaffolds while Unity fingerprints were used in the diversity optimization of the vasicinone library. The measured solubility, permeability in MDCK cells and CYP inhibition were measured for the scaffolds. The resulting libraries demonstrated favorable distributions of each of the measured ADME properties. 

While most reports in the literature employ a modified version of Lipinski s Rule-of-Five criteria, a few have used models that are more sophisticated as part of the design. Many QSAR models have been reported in the literature for many different ADMET related properties, from solubility to permeability to hERG inhibition to metabolic stability. Several recent 

The predictive models included in the program Volsurf were utilized as part of a library designed to identify non-benzodiazepine ligands for the GABA-A receptor. A virtual library of 500 compounds was enumerated using Accord for Excel. The virtual library was filtered to remove compounds 

Once identified, these functional groups can be optionally excluded from future compound acquisitions or library designs unless a competing interest encourages their use. A report on the compound acquisition process at Novartis discusses how compounds are prioritized or eliminated for acquisition based on similarity to the historical collection, as well as physicochemical properties and undesirable functionality. Many of the same themes are presented in other reports on compound selection.  

---