Study Finds ACD/Labs pKa Predictions to be Most Accurate

A recently published study comparing pKa prediction tools found ACD/pKa to be the best based on several criteria.

Toronto, Canada (March 11, 2007)—Researchers at Pardubice University in the Czech Republic published a study in the December 2007 issue of Analytical and Bioanalytical Chemistry evaluating the accuracy and predictive power of four pKa prediction software tools (including commercially available and free packages). Their conclusion, after in-depth statistical analysis, was that ACD/pKa DB provides the most accurate predictions.

Modern chemistry relies heavily on the modeling of chemical and biological parameters, based on the prediction of molecular properties. Not least because in-silico predictions help minimize the cost associated with experimental measurement. While tools for the prediction of a variety of molecular properties are available, prediction of one property in particular, pKa—measuring the degree of dissociation of a compound in solution—was chosen by the researchers due to its impact on pharmacology, and the effect of chemicals in the environment—to name but two areas.

The study, carried out by M. Meloun and S. Bordovska and entitled 'Benchmarking and validating algorithms that estimate pKa values of drugs based on their molecular structure' (Anal. Bioanal. Chem., 389: 1267–1291, 2007), compares predicted values to experimental results for three sets of compounds. Using a linear regression analysis technique (fully defined in the paper) they concluded that ACD/pKa DB provided the most accurate predicted values compared to the other packages, showing the best results in six different statistical characteristics.

All four of the software packages compared use of a fragment-based approach to prediction, meaning that the software consists of a database of chemical structure fragments, and uses a proprietary algorithm to calculate predicted values for whole molecules based on the fragments it contains. Therefore when comparing software, one must also take into consideration the size of the fragment database, and whether there is the ability to add one's own experimental data to the database, not to mention factors such as reliability, commitment of the manufacturer, and the availability of advanced features.

ACD/pKa DB uses Hammett equations derived from a library of ~14,000 highly curated compounds—one of the largest internal databases of any commercially-available pKa predictor—to predict aqueous pKa. ACD/pKa DB also offers two methods for adding proprietary experimental data to improve prediction accuracy for novel chemical space. In addition, two reference databases are available that offer quick look-up of published data—one contains >31,000 experimental pKa values for approximately 16,000 compounds in aqueous solutions; the other provides experimental data for more than 2000 molecules in non-aqueous solvents. Since the introduction of the first pKa predictor from ACD/Labs in 1995, there has been continual development and enhancement of the software which is currently used by the majority of pharmaceutical and API companies worldwide.

For more information about pKa prediction or ACD/Labs' Physicochemical prediction software, visit our website at

About Advanced Chemistry Development
Advanced Chemistry Development, Inc., (ACD/Labs) creates innovative software packages that aid chemical research scientists worldwide with spectroscopic validation of structures, elucidation of unknown substances, chromatographic separation, medicinal chemistry, preformulation of novel drug agents, systematic nomenclature generation, and chemical patenting and publication. Founded in 1994, and headquartered in Toronto, Canada, ACD/Labs employs a team of over 145 dedicated individuals whose continual efforts carry ACD/Labs' innovative technologies into pharmaceutical, biotech, chemical, and materials companies throughout the world. Information about Advanced Chemistry Development and its products is available at