Prediction of distribution coefficients from structure. Comparison of calculated and experimental data for various drugs.

The efficiency of the program PrologD to predict distribution coefficients (D) at any pH and pairing ion concentration has been tested using experimental logD values for various drugs measured under standard conditions of buffers and ionic strength. Clonidine derivatives, fluoroquinolones and beta-blockers were included as particular pharmacological classes within the testing data set. Calculations were performed using the three logP estimation options implemented in the program. PrologD proved to be very efficient and can be of great advantage in drug research. Prediction patterns and correlations between experimental and calculated data indicate acceptable results for more than 80% of the data. In addition, comparable studies using the different options permitted suggestions for the more suitable logP estimation method in respect of the particular classes of compounds.