Binding affinity data of twenty seven 1,2,4-triazolo[1,5-α]quinoxaline derivatives for human adenosine A₃ receptor subtype have been subjected to QSAR analysis using genetic algorithm (GA) method. In order to obtain perfect statistical quality model, descriptors from topological, thermodynamic, spatial, electrotopological and quantum chemical class were calculated. The best model is shown below and its index in statistially is LOF＝0.291, r²＝0.766, r_adj²＝0.723, F-test＝17.974, PRESS＝3.469, CV-r²＝0.791; pK_i＝13.407－0.027*F_C－8^E－0.033*F_C－8^N＋0.845*Atype_C_28－19.493*Shadow_XYfrac. Some conclusions are presented that the binding affinity of these compounds increases with reducing weighted electrophilic and nucleophilic atomic frontier electron density at the C-8 position, the fraction of area of molecular shadow in the XY plane over area of enclosing rectangle, and increasing the hydrophobic atom type descriptor (Atpye_C_28). According to this model, we also design two new compounds and predict that they may have high binding affinities. We expect that the present study provide theoretical instruction on the structural modification of quinoxaline derivatives as human A3 adeno-sine receptor antagonists, and may be helpful for probing the mechanism of action between the receptor and its antagonists.

Key words: genetic algorithm, binding affinity, quantum chemical descriptor, QSAR