Linear and non-linear QSAR models on platinum (II) anticancer drugs with N-donor ligands

This research presents a quantitative structure-activity relationship (QSAR) of IC50 values of Platinum(II) derivatives. Twenty one different platinum(II) anticancer derivatives have been selected as a sample set and the geometry of the complexes are optimized using Gaussian 03W. The activity of the 21 different Platinum(II) derivatives is estimated by means of multiple linear regression (MLR), artificial neural network (ANN), simulated annealing (SA) and genetic algorithm (GA) techniques. These methods are also utilized to select the most efficient subsets of descriptors in a cross validation procedure for non-linear log(IC50) prediction. The results obtained using the GA-ANN have been compared with those obtained using MLR-MLR, MLR-ANN, SA-ANN and GA-ANN approaches. A high predictive ability has been observed for the MLR-MLR, MLR-ANN, SA-ANN, MLR-GA and GA-ANN models, with root mean sum square errors (RMSE) of 0.127, 0.013, 0.011, 0.0125 and 0.0099, respectively (N=21). The results obtained using the GA-ANN method indicate that the activity of the derivatives of Platinum complexes depends on different parameters such as Mor9v, RDF140v and G2e descriptors. In summary, a comparison of the quality of ANN with different MLR methods shows that ANN has a better predictive ability.