Virtual library screening

For the generation of targeted libraries, similarity algorithms are useful for computationally screening virtual libraries. Since the number of structures available from such libraries is beyond practical limits, the tools and strategies to be applied for computational screening often differ from the ones used for physically available... 

Several research groups have built models using theoretical desaiptors calculated only from the molecular structure. This approach has been proven to be particularly successful for the prediction of solubility without the need for descriptors of experimental data. Thus, it is also suitable for virtual data screening and library design. The descriptors include 2D (two-dimensional, or topological) descriptors, and 3D (three-dimensional, or geometric) descriptors, as well as electronic descriptors. 

PRO SELECT One of the first reported tools for the virtual screening of libraries for fit into a protein active site Protherics Molecular Design Ltd. http //www.protherics. com [30]... 

Cathepsin D. The design of inhibitors of the aspartyl protease cathepsin D started from a virtual library of peptide analogs that contained the typical hydroxyethylamine isoster for the cleavable peptide bond. As the availability of starting materials would have generated a library of about 1 billion compounds, virtual screening was applied to reduce this multitude of candidate structures to a reasonable number. The backbone of a peptide... 

Wang RX, Wang SM. How does consensus scoring work for virtual library screening An idealized computer experiment. J Chem Inf Comput Sci 2001 41 1422-6. 

As stated above, an advantage of virtual screening is that any compound, real or virtual, can be screened and the user is not restricted to those compounds available in corporate or external collections. The technology can also be used to screen proposed libraries and even select monomers for a combinatorial library based on 3D fit to the target structure. 

Test a substantial number of compounds. VS methods generally offer enrichment, but most ranked hit lists contain a significant proportion of false positives. Hitlists should be scaled to 1-5% of the compounds in the virtual library screened. In many real world situations, the computational chemist is being asked to choose lists of compounds representing 0.1% or less of the compounds screened (e.g., the best 100 of 100,000 compounds). Typically, VS methods have been validated considering 1%, 5%, or 10% of the total number of compounds in the VS collection. By following up on more compounds, one increases the probability of impact from VS. 

A major drawback of the PSAa is, however, the rather time-consuming calculation, particularly, the Monte Carlo conformational search, which makes PSAa inappropriate for computational screening (e-screening) of large virtual libraries. 

In view of latter developments (see Sections 16.4.9-16.4.11 for further details) the procedure, even with simplifications such as using a single CONCORD/ CORINA-derived 3D geometry instead of performing a Monte Carlo conformational search, is too computationally expensive to be applied to e-screening of virtual libraries. However, it may still be a useful alternative/complement for computing more detailed information about a compound, or to provide a more easily interpretable model to complement other models based on more rapidly computable parameters but which are difficult to interpret in terms of how to modify compounds in order for them to have better intestinal absorption characteristics. 

The next vague of tools will be around computational or in silico ADME approaches. These will allow to include ADME into the design of combinatorial libraries, the evaluation of virtual libraries, as well as in selecting the most promising compounds to go through a battery of in vitro screens, possibly even replacing some of these experimental screens. Several of these computational tools are currently under development as will be discussed in this volume. 

An approach we term Virtual Screening of Virtual Libraries (VSVL) is intended to take advantage of our in-house automated synthesis capabilities. This involves the creation of large 3-D databases of virtual libraries using the Catalyst system [20]. Catalyst uses a prestored set of conformations for each molecule, so while construction of the... 

---