Anal. Bioanal. Chem., 389: 1267–91, 2007

Benchmarking and validating algorithms that estimate pK_a values of drugs based on their molecular structure

M. Meloun and S. Bordovska

Abstract

The REGDIA regression diagnostics algorithm in S-Plus is introduced in order to examine the accuracy of pK_a predictions made with four updated programs: PALLAS, MARVIN, ACD/pK_a and SPARC. This report reviews the current status of computational tools for predicting the pK_a values of organic drug-like compounds. Outlier predicted pK_a values correspond to molecules that are poorly characterized by the pK_a prediction program concerned. The statistical detection of outliers can fail because of masking and swamping effects. The Williams graph was selected to give the most reliable detection of outliers. Six statistical characteristics (F _exp, R², , MEP, AIC, and s(e) in pK_a units) of the results obtained when four selected pK_a prediction algorithms were applied to three datasets were examined. The highest values of F_exp, R², , the lowest values of MEP and s(e), and the most negative AIC were found using the ACD/pK_a algorithm for pK_a prediction, so this algorithm achieves the best predictive power and the most accurate results. The proposed accuracy test performed by the REGDIA program can also be applied to test the accuracy of other predicted values, such as logP, logD, aqueous solubility or certain physicochemical properties of drug molecules.

Download the article from the Journal of Analytical & Bioanalytical Chemistry web site.