|
KMID : 0043320170400020214
|
|
Archives of Pharmacal Research
2017 Volume.40 No. 2 p.214 ~ p.230
|
|
|
A new model of flavonoids affinity towards P-glycoprotein: genetic algorithm-support vector machine with features selected by a modified particle swarm optimization algorithm
|
|
Cui Ying
Chen Qing Gang
Li Ya Xiao
Tang Ling
|
|
|
Abstract
|
|
|
|
Flavonoids exhibit a high affinity for the purified cytosolic NBD (C-terminal nucleotide-binding domain) of P-glycoprotein (P-gp). To explore the affinity of flavonoids for P-gp, quantitative structure?activity relationship (QSAR) models were developed using support vector machines (SVMs). A novel method coupling a modified particle swarm optimization algorithm with random mutation strategy and a genetic algorithm coupled with SVM was proposed to simultaneously optimize the kernel parameters of SVM and determine the subset of optimized features for the first time. Using DRAGON descriptors to represent compounds for QSAR, three subsets (training, prediction and external validation set) derived from the dataset were employed to investigate QSAR. With excluding of the outlier, the correlation coefficient (R2) of the whole training set (training and prediction) was 0.924, and the R2 of the external validation set was 0.941. The root-mean-square error (RMSE) of the whole training set was 0.0588; the RMSE of the cross-validation of the external validation set was 0.0443. The mean Q2 value of leave-many-out cross-validation was 0.824. With more informations from results of randomization analysis and applicability domain, the proposed model is of good predictive ability, stability.
|
|
|
KEYWORD
|
|
|
Flavonoids, Quantitative structure?activity relationships, Modified particle swarm optimization, Genetic algorithm-support vector machine, P-glycoprotein affinity
|
|
|
FullTexts / Linksout information
|
|
|
|
|
|
Listed journal information
|
|
  
|