Computer prediction of biological activity spectra as a tool for determining the priorities in biological testing

Biological activity is one of the most important characteristics of a chemical compound reflecting its interaction with living organisms. Since the majority of chemical compounds have many different types of biological activity, complex evaluation of their biological potential is necessary. Such evaluation can be done via the Internet with the computer program PASS [1] which estimates the probabilities for 900 types of biological activity on the basis of structural formulas of compounds with average accuracy of ~85%. PASS predictions are based on the analysis of structure-activity relationships (SAR) for the training set consisting of about 46000 biologically active compounds. Despite the incompleteness of the training set, the PASS algorithm is robust enough to obtain a reliable SAR [2] or even to predict a compound’s "drug-likeness" [3]. Calculation of biological activity spectra for 10,000 compounds on an ordinary PC takes several minutes, thus PASS can be effectively used to determine the priorities in testing both for a particular compound, a few ones or many thousands compounds from large databases. In the first case the most probable activities will be studied in particular compounds [4], in the second case the most prospective compounds will be selected among the samples available for testing [5, 6]. Application of PASS to the diverse set of drug-candidates reveals some new biological activities in these compounds [4]. It was shown that due to the PASS prediction, the population of “actives” can be enriched up to 17 times [5]. Based on the PASS prediction, new antihypertensive compounds with dual mechanism of action were found [6]. Since the number of known pharmacological targets will be increased from ~500 to ~5000-10000 in a few years due to the achievements of genomics, proteomics and bioinformatics, complex evaluation of chemical compounds will become even more important. PASS is open for adding new compounds and biological activities to the training set when such novel data will be discovered. State-of-the-art and future trends in computer-aided prediction of biological activity spectra will be discussed.

1. PASS: Prediction of Activity Spectra for Substances. Available from URL: http://www.ibmh.msk.su/PASS.
2. Poroikov V., et al. Robustness of biological activity spectra predicting by computer program PASS for non-congeneric sets of chemical compounds. J. Chem. Inform. Comput. Sci., 2000; 40: 1349-1355.
3. Anzali S., et al. Discriminating between drugs and nondrugs by Prediction of Activity Spectra for Substances (PASS) J. Med. Chem., 2001; 44: 2432-2437.
4. Stepanchikova A.V., et al. Prediction of biological activity spectra for substances: Evaluation on the diverse set of drugs-like structures. Current Med. Chem., 2003, 10: 225-233.
5. Poroikov V.V., et al. PASS Biological Activity Spectrum Predictions in the Enhanced Open NCI Database Browser. J. Chem. Inform. Comput. Sci., 2003, 43: 228-236.
6. Lagunin A.A., et al. Computer-aided selection of potential antihypertensive compounds with dual mechanisms of action. J. Med. Chem., 2003. 46 (15), 3326-3332.