Figure 22. Classification of the NSAID dataset based on three-dimensional autocorrelation descriptors: Nonlinear map. The regions marked indicates classes of compounds. The dark gray region displays COX-2-selective drugs.
The simple NSAID example shows which decisions and parameters influence the result of similarity or diversity examinations. Most important is the choice of the descriptors set which defines the molecular properties considered. The method used for classification and any further computation has only a small effect on the result.
Topological descriptors can be computed very quickly for a huge number of compounds. These descriptors are able to reproduce the intuitive classification of structural formulas. For this reason, topological descriptors are a good choice to describe diversity of huge virtual libraries in an automated way.
Classification based on three-dimensional molecular structures requires more time-consuming computations, but results in more reliable information. These descriptors are able to identify different pharmacological profiles of compounds and thus provide the chemist with novel information that is often not recognizable from the structural formulas directly.
Was this article helpful?