ABSTRACT
The method of relating structural properties with biological activity by means of quantitative structure-activity relationships (QSAR) is increasingly being applied to toxicological effects—allergic contact sensitization among them. A biologically based QSAR model discriminating between allergens and nonallergens was developed and tested successfully in diverse studies. The process of QSAR model construction involves collection of pertinent clinical data and analyzing the database thus obtained for statistically significant parameters by discriminant or regression analysis from among a wealth of chemical and physicochemical descriptors. A further step in developing a model that is truly predictive of dermal sensitization potential is the establishment of a rank order in the sensitization potency of chemicals. In vivo validation of the classification model was accomplished by matching its prediction against sensitization data obtained in the mouse. The difference between the classification model and the ranking model lies only in the descriptor coefficients for the algorithm.
