From knowledge generation to knowledge archive. A general strategy using TOPS-MODE with DEREK to formulate new alerts for skin sensitization.

A general strategy for knowledge flow concerning skin sensitization based on the combined use of TOPS-MODE and DEREK expert system is proposed. TOPS-MODE is used as a knowledge generator, while DEREK represents the knowledge archive. A TOPS-MODE classification model allows the identification of structural fragments and groups responsible for strong/moderate skin sensitization. These structural contributions are sorted, analyzed, and graphically displayed in an appropriate way allowing the identification of several structural alerts for skin sensitization. Nine structural alerts already implemented in DEREK are identified using this strategy. They comprise, among others, alkyl halides, aldehydes, alpha,beta-unsaturated compounds, aromatic amines, phenols, hydroquinone, isothiazolinone, and alkyl sulfonates. Four new hypotheses are generated using TOPS-MODE structural contributions to skin sensitization, which are not recognized as structural alerts by DEREK. They include the reduction of aromatic nitro groups and epoxidation reaction of double bonds as metabolic activation steps that can lead to reactive haptens which can trigger the skin sensitization mechanism. Another new alert is based on 1,2,5-thiadiazole-1,1-dioxide for which we have identified a possible mechanism explaining its strong skin sensitization profile. It is based on the existence of a tautomeric equilibrium and further reaction with nucleophiles, which are both supported by experimental evidence. Finally, we have identified a possible new mechanism for the skin sensitization of nonreactive compounds, which involves the formation of noncovalent complexes with proteins in a processing- and metabolism-independent way.