Optimization of experiments in combinatorial chemistry

synthesis of a complete combinatorial library in all its variability Is expensive, undesirable or even impossible. In Subsection 5.3.6 we considered exeimples where between 10 and 20 reactants resulted in several thousand different compounds. Thus, often only a subset of the virtual library is synthesized, termed the real library. The virtual library consists of all compounds available when using all possible combinations. 

Initially, the structures of the real library are mapped onto real numbers using molecular descriptors. A vector of equal length of real numbers is obtained for each structure. Further, an experimental value of the property of interest is associated with each structure. Descriptor and property values are the input for statistical methods of supervised learning. These result in a predicting function, a function that is determined to lit the experimental property values reasonably weD (see Chapter 6). The predicting function requires a vector of descriptor values as input and returns a predicted property value as output. The complete process of descriptor calculation, the search for and hnally application of a predicting function is known as QSPR/QSAR research. 

The virtual library is constructed using a structure generator. Of course, descriptors can be calculated for each structure in this library. Therefore the predicting function is able to predict property values for each member of the virtual Ubrary. This is called virtual screening. Structures predicted to exhibit desired property values are candidates for targeted synthesis. 

---