Applications of Genetic Algorithms on the Structure-Activity Relationship Analysis of Some Cinnamamides
暂无分享,去创建一个
Quantitative structure-activity relationships (QSARs) for 35 cinnamamides were studied. By using a genetic algorithm (GA), a group of multiple regression models with high fitness scores was generated. From the statistical analyses of the descriptors used in the evolution procedure, the principal features affecting the anticonvulsant activity were found. The significant descriptors include the partition coefficient, the molar refraction, the Hammet sigma constant of the substituents on the benzene ring, and the formation energy of the molecules. It could be found that the steric complementarity and the hydrophobic interaction between the inhibitors and the receptor were very important to the biological activity, while the contribution of the electronic effect was not so obvious. Moreover, by construction of the spline models for these four principal descriptors, the effective range for each descriptor was identified.
[1] R. Leardi. Application of a genetic algorithm to feature selection under full validation conditions and to outlier detection , 1994 .
[2] A. J. Hopfinger,et al. Conformational Properties of Macromolecules , 1973 .
[3] R. Boggia,et al. Genetic algorithms as a strategy for feature selection , 1992 .