Conformational databases

J Kuszewski, AM Gronenborn, CM Clore. Improving the quality of NMR and crystallographic protein stiaictures by means of a conformational database potential derived from structure databases. Protein Sci 5 1067-1080, 1996. 

Fig. 12 The 2Fob-Fcak omit map used in the automated query. The 2Fobs-Fca c omit map calculated with all reflections to 2.88 A using CNS, map was phased from five shaken high temperature annealed protein models also run at 2.88 A. This omit map exhibits a similar spatial distribution and total volume by comparison with the map in Fig. 11a, but it possesses a single contiguous volume suitable for automated fitting of conformational databases as described in the text. X-ligand was used to locate maximum unoccupied volume of density (highlighted in gold) and to flexibly fit epothilone rotamers from the conformational database. Only two orientations of ligand, as depicted above, were found to satisfy the density shape...

So, multiple MD is a good tool to assess the stability of conformers short biomolecules (this was also seen for the related peptide Tyr-Thr-Gly-Pro [50]). Due to the long length of the equilibration phase of a biomolecular system in aqueous solution, multiple MD simulations cannot be used in a blind systematic manner on a large ensemble of conformations. But it is very informative on a selected set of conformers, which have been derived by knowledge based conformational searches. The limit of the multiple simulation size in explicit water is about 10 conformers. Database searches seen to be a very good tool to derive low energy conformers all the conformations found in the structural database were found to be at least 1% populated in the 33.6 ns of simulation. 

Application Multiconformer database Single-conformer database... 

There is one other special case which should be mentioned the pre-calculated multi-conformer database. As we shall see later, a sizable fraction of the CPU time in evaluating a 3D virtual library can be taken up with the generation of conformers. However, let us assume that we have access to a small in-house sample collection database with 100000 compounds. If we are willing to limit ourselves to studying the compounds in this library, we can calculate a multi-conformer database with perhaps 10 to 50 conformers per molecule. Since the database is static, the conformers need only be generated once, stored, and accessed as needed. 

Once a virtual library has been created and the undesirable compounds have been removed, a natural next step is to generate one or more 3D conformations for each molecule. Several recent studies have shown that multi-conformation databases are more effective for docking and 3D searching. " If the virtual screening method is to evaluate flexible molecules, then the generation of multiple conformers is essential. 

---