Multivariate image – quantitative structure activity relationships of the antibacterial activity of pleuromutilin derivatives using different chemometrics methods

Multivariate image analysis applied to quantitative structure-retention relationship (MIA-QSRR) has shown to be a useful tool to model the antibacterial activity of pleuromutilin derivatives. Antibacterial activity of 55 pleuromutilin derivatives used in this study was reported by Hirokawa et al. [1]. In MIA-QSPR [2], images are twodimensional chemical structures, such as those drawn by using known programs like ChemDraw. These images (2D chemical structures) have shown excellent correlation with retention times and are supposed to codify chemical properties, like size of substituents, chains, branches and chiral centers. Chemical structure of pleuromutilin derivatives In order to achieve this, different methods were employed in this study: partial least squares (PLS), genetic algorithm-PLS (GA-PLS), orthogonal signal correction-PLS (OSC-PLS) and PC-least squares-support vector analysis (PC-LSSVM) [3]. The results of all models are compared with statistical parameters such as RMSEP, RSEP, R2 and Q2. The resulted model showed high prediction ability with root mean square error of prediction of 0.0062 and 0.0024 for OSC-PLS and PC-LSSVM. Results have shown that the introduction of PC-LSSVM for pixel descriptors drastically enhances the ability of prediction in QSAR studies superior to other calibration algorithms.