For decades pharmaceutical companies has been spending millions of UAH on research and development of new types of drugs. Way the therapeutic compound must went through to become available to end consumer consists of many stages including computer modeling which is the easiest, cheapest and readily available for everyone. It is estimated that every year in U. S. only over 26 million animals are being used by pharmaceutical companies for drug testing and much more all around the globe. This number can be reduced by using computer methods for predicting effects of drugs before animal testing.

Purpose: to evaluate GUSAR and PASS methods as a first stage of predicting active and adverse effects of hypoglycemic drugs which act on PPARs. And to determine if it’s a reliable method for drug research. Also, this will allow to reduce number of laboratory animals used for testing and discovering active effects.

Materials and methods: Toxicity was predicted using GUSAR with QSAR models (Quantitative structure–activity relationship) and PASS program to prognosticate active and toxic therapeutic effects. Compounds that were tested: Thiazolidinediones which are used in treatment of Type 2 diabetes.

Results: Programs successfully predicted active and toxic effects of both low molecular mass drugs which then were confirmed by clinical trials. Predicted effects for malabsorbable preparations cannot be clinically proved because of its low bioavailability.

Conclusion: methods based on QSAR analysis like GUSAR and PASS are good start point for drugs development, but they need further development and evaluation. Using suggested programs can lower cost and time needed for drug development. It will greatly reduce number of laboratory animals needed for drug testing by providing possible side effects before trials.