In this article, the R^2 guided region selection CoMFA method was used to investigate 12 saponins for their 3D-QSAR. The impact of the probes on the result of CoMFA model is analyzed including some complex probe models. The final result of CoMFA model proved that the contribution to activity from steric field is 40%, and that from electrostatic field is 40%. The cross-validated R^2 of the CoMFA model is 0.653, and the non-crossvalidated R^2 is 0.991. The variance ratio F(4,7) obtained is 130.195 with confidence level above 99%, and the ratio of standard error and extremal length obtained, s/△γ, is 4.2%. These results proved that the CoMFA model has good prediction, and some new compounds with higher activity can be disigned based on it.

Key words: SAPONINS, THREE DIMENSIONAL STRUCTURE, QUANTITATIVE STRUCTURE ACTIVITY RELATIONSHIP