Modeling and prediction (correction) of partition coefficients of bile acids and their derivatives by multivariate regression methods

Different multiple regression methods including forward stepwise multiple linear regression (MLR), principal component regression (PCR) and partial least squares (PLS) have been applied to the modeling of partition coefficient (lipophilicity) of bile acids and their derivatives by means of 16 different descriptors obtained by using Alchemy package software and retention index RMo as an experimental estimation of lipophilicity. Retention indices for bile acids and their derivatives were determined by reversed phase high-performance thin layer chromatography on RP-18W bounded stationary phase with methanol–water in different volume proportions as mobile phase. The results achieved concerning the prediction of Log P are highly significant and consistent with the molecular structure of the compounds investigated. The sum of absolute values of the charges on each atom of the molecule, in electrons (SQ), the sum of absolute values of the charges on the nitrogens and oxygens in the molecule, in electrons (SQNO), specific polarizability of a molecule (SP), the third-order connectivity index (3χ) and molecular lipophilicity, seem to be dominant in the partition mechanism. In addition, regression models developed have allowed a correct estimation of the partition coefficients of cholic acid (Log PHA = 2.93; Log PA−=2.02) as compared with reported experimental values (Log PHA = 2.02; Log PA−=1.1).