Structure-retention relationship study of arylpiperazines by linear multivariate modeling

Abstract

A quantitative structure retention relationship study has been performed to correlate the retention of 33 newly synthesized arylpiperazines with their molecular characteristics, using thin-layer chromatography. Principal component analysis followed by multiple linear regression (MLR), principal component regression (PCR) and partial least squares (PLS) was performed to identify the most important factors, to quantify their influences, and to select descriptors that best describe the behavior of the compounds investigated. The best statistical performance was achieved by applying PLS regression, leading to the lowest value of the standard error (root mean square errors of calibration of 0.159 and cross-validated value RMSE cross-validation = 0.231 units), followed by the PCR (root mean square errors of calibration = 0.195 and RMSE cross-validation = 0.305) and MLR (R(adj)(2) = 0 9499, F = 102.017, mean square error = 0.052 and predicted residual error sum of squares = 2.23). Two factors of the highest influence: surface tension and hydrophilic lipophilic balance appear as the part of obtained models. In addition, polar surface area and hydrophilic surface area are included by both PLS and PCR models. Moreover, log P has been added to the PLS model. Besides, PCR model includes following descriptors: hydrogen bond acceptor, hydrogen bond donor and LUMO energy, whereas topological descriptors: connectivity indices 0 and 2, and valence index 3 are included in the MLR model.