Diversity has historically played a critical role in design of combinatorial libraries, screening sets and corporate collections for lead discovery. Large library design in the 1990’s ranged from arbitrary through property based reagent selection to product based approaches. Over time, however, there has been a downward trend in library size as information about the desired targets increased due to the genomics revolution and the increasing availability of target protein structures from crystallography and homology modeling. Concurrently, computing GRIDs and CPU clusters have facilitated the development of structure based tools that screen hundreds of thousands of molecules. Smaller 'smarter' combinatorial and focused parallel libraries have replaced those un-focused large libraries in the twenty-first century drug design paradigm. While diversity still plays a role in lead discovery, target family and target specific approaches dominate current efforts in library design. This talk will highlight these trends and explore some of the recent structure based approaches.

During the discovery of the first oral drug that targets influenza neuraminidase (NA), we discovered a novel nonpolar interaction between an NA binding pocket and a potent inhibitor GS4071 (Oseltamivir). This apparently highly polar pocket interacts favorably with the hydrophobic alkyl moiety of GS4071, this unusual interaction plays an essential role in determining the potency of GS4071. We hypothesize that the pi-faces of the sp2 planar termini of these polar pocket residues may behave similarly to those of aromatic rings and therefore could interact favorably with hydrophobic moieties of a bound inhibitor. Given the progress made in virtual screening technologies since then, a retrospective study was conducted utilizing these newer methods. The objective here is to determine whether virtual screening would have added value towards the discovery of GS4071. In addition, we want to find out whether or not today's scoring functions can sufficiently take this novel interaction into consideration.

An important problem in virtual screening is the quality of the molecular representations being screened. Careful Database creation and curation are ongoing problems for experts in the field and barriers to entry for non-experts. This has led us to create ZINC, a free database of commercially available compounds for virtual screening, on the web at http://zinc.docking.org. ZINC contains multiple representations of molecules, since molecules that are not in the biologically relevant protonation state, tautomeric, stereo- or regio-isomeric form in the virtual library often fail to dock and score well. ZINC has been broadly sub-categorized into several libraries, such as 'drug like', 'lead-like' and 'fragment-like' molecules, for different applications that we ourselves at UCSF find useful. Several of these will be illustrated by case studies. Users may also create their own focused library subsets via a web-interface. The exploration of ZINC by other groups in both industrial and academic laboratories will be briefly discussed.