Representative training

Our philosophy was instead top-down . We decomposed the molecular TAE (total atomization energy TAEe at the bottom of the well, TAEo at absolute zero) into all components that can reasonably affect it at the kl/mol level. Then we carried out exhaustive benchmark calculations on each component separately for a representative training set of molecules. Finally, for each component separately, we progressively introduced approximations up to the point where reproduction of that particular component started deteriorating to an unacceptable extent. Thus, experimental data entered the picture only at the validation stage, not at the design stage. 

Holtje and Sippl selected a representative training set of twelve superimposed structures. The aforementioned Y AK process resulted in a pseudoreceptor that consists of six amino acid residues (Figure 5). In this pseudoreceptor model, the imidazole ring of the ligands is bound to a tyrosine residue that donates a proton to the imidazole and an asparagine residue that accepts a proton from the imidazole. Hence, two distinct H-bond have been found, which explain the strict binding mode of the imidazole. It has to be noted that the position of these receptor residues relative towards the imidazole is a consequence of... 

Molecular supergraph for a series of tetrahydroimidazobenzodiazepinone (TIBO) derivatives with superimposed structure of a representative training set compound. 

Figure 5.9 Molecular supergraph for a series of melatonin receptor ligands with superimposed structures of several representative training set compounds.

Fig. 5.30. Representative trains of oscillatory glow for CO + O2 mixtures in a closed system (a) chloropicrin-treated reactor with p = 33 Torr, = 841 K giving a total of 94 oscillations over 3 hr (b) and (c) light emission and reactant consumption in a clean reactor with p = 22 Torr and T, = 900 K. (Reprinted with permission from reference [60], Royal Society...

The QSAR is derived with a set of chemicals presenting the entire range of compounds concerned (the representative training set). 

Selection of a representative training set of compounds covering the entire domain of relevant properties. 

The working model considers existence of train, inspection and maintenance. The place TR busy represents a track occupied by train, the place TR idl represents the component being available. Transition T D means that a train enters this track component. T E represents train leaving. Substitution transitions TRV and VI represent the inspection arrivals. Substitution transition Cl combines the component deteriorations and the maintenances. 

Two types of pattern recognition technique " have been applied to spectrum interpretation. Supervised methods (see Supervised Learning) are limited to one or more predefined structural classes and require representative training sets for each to develop the classifier. Unsupervised methods (see Unsupervised Learning) partition a set of spectra into clusters with common structural features on the basis of spectral features alone. No predefined classes are required. 

---