Quantitative structure activity relationships (QSAR) of anti-cancer isatin derivatives were discovered by multiple linear regressions (MLR) and genetic algorithm partial least squares (GA-PLS) methods. Topological, chemical, geometrical and functional groups descriptors were found to be effective parameters on the cytotoxic activity. The positive effects of the number of halogen atoms and the number of total secondary carbons, and the negative effects of the number of secondary amides, and the number of ketones on the anti-cancer activity were in agreement with previous SAR studies. Hansch analysis showed the importance of lipophilic R(3) and R(5) substituents. Between MLR and GA-PLS, MLR represented superior results with a high statistical quality (R(2)=0.92 and Q(2)=0.90) for predicting the activity of the compounds.Copyright (c) 2009 Elsevier Masson SAS. All rights reserved.
