TOXICITY MODELLING OF SOME ACTIVE COMPOUNDS AGAINST K562 CANCER CELL LINE USING GENETIC ALGORITHM-MULTIPLE LINEAR REGRESSIONS

Authors : David Ebuka Arthur

Pages : 355-374

Doi:10.18596/jotcsa.287335

View : 8 | Download : 6

Publication Date : 2017-01-09

Article Type : Research Paper

Abstract :This research entails the modelling of the toxicity of anticancer compounds on K562 cell line, where 112 compounds that make up the data set were divided into training and test set to be used for developing and validating the model respectively. The internal and external validation parameter R 2 for the training and test set given as 0.845 and 0.5316 respectively justifies the robustness and the ability of the model to predict toxicity of the compounds. WPSA-3 and minHBint7 molecular descriptor is responsible for about 50% of the overall effect on the model.
Keywords : QSAR, Model, External Validation, Molecular descriptors, Genetic algorithm