Knowledge-based descriptors

Nissink, J.W.M., Verdonk, M.L., and Klebe, G. Simple knowledge-based descriptors to predict protein-ligand interactions. Methodology and validation. /. Comput.-Aided Mol. Des. 2000, 14, 787-803. 

METEOR S biotransformation rules are generic reaction descriptors, and the versatile structural representation used in the system allows each atom or bond to have specific physicochemical properties. This approach provides more details than simple hard-coded functional group descriptors (313), but this flexibility also can give rise to an avalanche of data. METEOR manages the amount of data by predicting which metabolites are to be formed rather than all the possible outcomes (310,312,314,315). At high certainty levels, when chosen, only the more likely biotransformations are requested. At lower likelihood levels, the more common metabolites are also selected for examination. Currently, METEOR knowledge-based biotransformations are exclusively for mammalian biotransformations (phase I and phase II) (314,315). 

Nowadays, more than 4000 types of descriptors are known.17 There exist different ways to classify them. With respect to the type of molecular representation used for their calculations—chemical formula, molecular graph, or spatial positions of atoms—one speaks about ID, 2D, and 3D descriptors, respectively. Descriptors can be global (describing the molecule as a whole) and local (only selected parts are considered). One could distinguish information-based descriptors, which tend to code the information stored in molecular structures, and knowledge-based (or semiempir-ical) descriptors issued from the consideration of the mechanism of action. Most of those descriptors can be obtained with the DRAGON, CODESSA PRO, and ISIDA programs. 

A separate class of experimental evaluation methods uses biological mechanisms. An artificial neural net (ANN) copies the process in the brain, especially its layered structure and its network of synapses. On a very basic level such a network can learn rules, for example, the relations between activity and component ratio or process parameters. An evolutionary strategy has been proposed by Miro-datos et al. [97] (see also Chapter 10 for related work). They combined a genetic algorithm with a knowledge-based system and added descriptors such as the catalyst pore size, the atomic or crystal ionic radius and electronegativity. This strategy enabled a reduction of the number of materials necessary for a study. 

The approach illustrated herein, exemplified by MULTICASE (3), is knowledge based. The input consists of the structures and toxicological activities of the chemicals in the learning set. The program then identifies structural descriptors (toxicophores) that are significantly associated with activity (see Table 19.3). The human expert participates in setting criteria for the inclusion of experimen-... 

In contrast, the emerging theme in knowledge based modelling is an ever closer intertwining of modelling and experimentation. A powerful combination of three factors contribute to this theme (i) structure-property relationships that are based on better models and more relevant structural descriptors (ii) robotics techniques in experimentation, which lead to a real explosion in the amount of data available (iii) informatics tools, which can handle the large amounts of data generated by both experiment and simulation. 

Coming back to the issue raised in the introduction physical modelling to replace experiment and knowledge-based modelling to couple experiment and simulation more closely the two opposing trends My conclusion is that they are in fact complementary improvements in the physical modelUng will lead to better models and descriptors for correlation based approaches, while more... 

RTXPS uses natural language syntax for descriptors, rules, and actions, supported by a script language for developing knowledge base. 

In using substructure-based descriptors, it is important to selectthe substructures to be considered carefully. Usually, for example in [290], a fixed set of substructures is given. This procedure is of limited use if compounds do not differ sufficiently with respect to these substructures, or if there are other substructures more suitable to express the structural differences in the compounds in question. In such a situation, user-defined substructures are helpful. This option is provided in the software MOLG EN-QSPR that resulted from this work. However, a prerequisite is good knowledge of the library, which requires some time and effort. This effort can be minimized using a computer. 

---