Enhancing the flexibility and adaptability of the DARC structural representation for computer-aided drug design.

Noticeable progress has been achieved in the determination of dynamic topochromatic variables for the structural representation of compounds and their situation in a given population. These independent structural variables can be further combined into more complex variables. Their contributions to the evolution of an associated property can therefore be evaluated with certainty. The risk of having correlated variables is avoided while the structural description remains exhaustive. In order to enhance the interpretative ability of the QSAR model, one or several physicochemical properties can be taken with these structural parameters as explanatory variables. Typically, partition coefficients, 3-D and quantum mechanical data are used for this purpose. The structural aspects not taken into account by the physicochemical parameters are reflected in the remaining topochromatic variables. The use of these new concepts is presented in a study of carbazole mutagenicity. The model explains 99% of the total variance with one external property and four additional topochromatic variables. The modulation of the heat of formation of an intermediate by two topochromatic variables suggests a much more precise interpretation than a simple combination of the usual external variables.