Pharmacological classification of drugs by principal component analysis (PCA) based on molecular modeling and high-performance liquid chromatography (HPLC) retention data is proposed. First, a group of 20 drugs of recognized pharmacological classification are chromatographed in eight diversified HPLC systems, applying columns with octadecylsilanes, phosphatidylcholine, as well as α1-glycoprotein and albumin. Additionally, molecular modeling studies, based on the structural formula of the drugs considered, are performed. Sixteen structural descriptors are derived. A matrix of 20 × 24 HPLC data together with molecular parameters are subjected to principal component analysis, and this revealed five main factors with eigenvalues higher than 1. The first principal component (factor 1) accounted for 47.8% of the variance in the data, and the second principal component (factor 2) explained 21.0% of data variance. The total data variance was 82.6% and is explained by the first three factors. The clustering of drugs is in accordance with their pharmacological classification, which proved that the PCA of the HPLC retention data, together with their structural descriptors, allowed the drugs to be segregated accurately to their pharmacological properties. This may be of help in reducing the number of biological assays needed in the development of a new drug.

The usage of principal component analysis (PCA) method in prediction of pharmacological classification of the drugs based on high-performance liquid chromatography (HPLC) retention data and on non-empirical structural parameters was studied. A group of 36 drugs of established pharmacological classification were chromatographed in ten carefully designed HPLC systems. Additionally, twelve structural descriptors were derived by molecular modeling studies based on the structural formula of considered drugs. A matrix of 36 x 22 HPLC data together with molecular properties parameters was subjected to chemometric analysis by PCA. Although that size of the training set could be sometimes disputable, the work remains as a demonstration of the basic methodology without the straight focus primarily intended asa report on a comprehensive predictive model. Nevertheless, the obtained clustering of drugs was in accordance with their pharmacological classification as well as chemical structures classification. The PCA method of the HPLC retention data and structural descriptors allowed to segregate drugs and drug candidates according to their pharmacological properties,and may be of potential help to limit the number of biological assays in the search for new drugs.

No abstract