Cardiac glycosides are group of natural products found to be effective against various cancers. The aim of this study is to propose a binding mode for different conformation of cardiac glycoside analogues to Na+, K+ - ATPase pump which is an important target for various cancer cell lines. Quantitative structure activity relationships model were developed with the cytotoxic activity (Expt. IC50) of 19 compounds based on molecular descriptors like docking score, binding free energy, ADME properties, eMBRAcE solvation model, and pharmacophore based 3D QSAR. In the cases of docking score and binding free energy, the correlation coefficient (R2) was in the range of 0.65–0.98 indicating good data fit, cross validation coefficient (q2) in the range of 0.64-0.99 and RMSE was in the range of 0.00–0.36 indicating that the predictive capabilities of the models were acceptable. The prediction model developed for the 226 conformers using docking score and binding free energy showed R2 in the range of 0.70-0.71, q2 in the range of 0.69-0.71 and RMSE in the range of 0.22-0.28 indicating acceptable prediction capabilities. In addition, a linear correlation was observed between the predicted and experimented pIC50 based on ADME properties with R2 of 0.88, and q2 of 0.73 and RMSE = 0.32. The prediction model developed for 226 conformers using ADME properties indicated a better fit with R2=0.99, q2= 0.99 and RMSE = 0.28. Further the prediction model developed using liaison showed, R2=0.90, q2=0.86, and RMSE=0.23; R2=0.59, q2=0.34 and RMSE= 0.45 for original and conformers respectively showing good to non linear response. The prediction model developed using eMBRAcE showed R2= 0.82, q2= 0.79 and RMSE= 0.28; R2= 0.89, q2= 0.89 and RMSE= 0.20 for original and conformers justify the solvation model parameters are good to use. Finally all the models were validate by pharmacophore based 3D QSAR model for the 19 cardiac glycosides in which the best hypothesis generated correlation coefficient, R2 =0.9733 which is acceptable pharmacophore features . Low level of RMSE and significant R2 and q2 values between Expt. IC50 and Predicted IC50 revealed that the best quality fit based upon all above said approach.
