Combinatorial DOCK

Bohm HJ, Banner DW, Weber L. Combinatorial docking and combinatorial chemistry design of potent non-peptide thrombin inhibitors. J Comput Aided Mol Design 1999 13 51-6. 

Bohm described combinatorial docking [272] implemented in LUDI [216] for reductive amination in search for thrombin inhibitors. The binding mode of the most active one with a Ki value of 95 nM was even confirmed by X-ray structure analysis [234] (see Figure 4.5e). 

The DOCK combinatorial docking implementation [275] was also applied for the design of novel enzyme inhibitors [276, 277]. In one of these prospective examples, Haque et al. reported potent low-nanomolar plasmepsin II aspartyl-protease inhibitors [278] from a set of several focused libraries with a best Ki value of 2 nM (cf Figure 4.5f). 

BOhm, H.-J. Combinatorial docking. In Computer-Assisted Lead Finding and Optimization, Van De waterbeemed, H., Testa, B., and Eolkers, G. (Eds.). Verlag Helvetica Chimica Acta, Basel, CH, 1997, 125-133. 

I. D. (1998) CombiDOCK structure-based combinatorial docking and library design. J Comput Aided Mol Des 12, 597-604. 

Combinatorial Docking Problem Given a library of ligands, calculate the docking score (and the geometry of the complex) for each molecule of the library. 

Methods for treating these problems have emerged from the area of molecular docking and de novo ligand design (see, e.g., Kubinyi158 for an overview on combinatorial docking methods). 

CombiDOCK Structure-Based Combinatorial Docking and Library Design. 

Combinatorial Docking and Combinatorial Chemistry Design of Potent Non-Peptide Thrombin Inhibitors. 

Although combiBUILD has been shown to be useful when the binding orientation of the scaffold is known, that is not always the case. The combinatorial DOCK algorithm has been developed as a more general tool for examining virtual libraries [52], Combinatorial DOCK places more emphasis on searching orientational space, that is how possible combinatorial compounds are oriented in the active site, and less emphasis on searching conformational space. 

Figure 17. The combinatorial DOCK algorithm. The receptor site is filled with spheres and then the scaffold atom-atom internal distances are matched to the site sphere-sphere distances. The matching process is used to orient the scaffold inside the active site. All components from all attachment sites are placed on the scaffold individually and scored. The top scoring components are combined on the scaffold, and tested for intramolecular clashes. Resulting best scores are saved. The process is repeated for a new orientation.

This approach was verified by performing a retrospective examination of the library designed by combiBUILD for cathepsin D. Since experimental results exist for all the library compounds, the question was how well could the combinatorial DOCK algorithm predict the active compounds from this set, compared to random. This was measured by looking at the enrichment factor or the ratio of active compounds predicted by DOCK compared with those expected from random predictions (see figure 18). For example, using 330nM to define active, DOCK finds approximately 78% of actives in the first 75 compounds (out of 1000 compounds total) and 87% of the actives in the top 200 compounds, for an overall enrichment factor of 10 [53],... 

The overall results show that the combinatorial DOCK can be useful in designing libraries, providing as much as an enrichment factor of 4. It should be noted that these results cannot be readily compared with those... 

Sun, Y., Ewing, T.J.A., Skillman, A.G. and Kuntz, I.D. Combinatorial DOCK Structure-based Combinatorial Docking and Library Design, J. Comput.-AidedMol. Des., 1998, in press. 

An alternative to sequential docking can be followed if combinatorial libraries are evaluated. Quite a few programs have been specifically designed for speed-up by so-called combinatorial docking. They profit from the structured, incremental nature of combinatorial libraries and the fact that molecules of a combinatorial library consist of a common core. This core is assumed to form common specific interactions with the receptor (possibly supported by experimental evidence) and can thus be prepositioned in the binding pocket in one or a few similar orientations. It then serves as skeleton for the addition of substituents. Obviously, this step is ideally suited for incremental construction algorithms (361)... 

Generally speaking, combinatorial docking approaches work best in cases where a core fragment plays a dominant role in the binding... 

---