, 1994; Waller et al., 1993). $$ r^ 2_\textpre = \left(\textSD-\textPRESS \right)/\textSD $$where SD is the sum of the squared deviations between the biological activities of molecules in the test set and the mean activity of the training-set molecules, and PRESS is the sum of the squared deviations between predicted and actual biological activity values for every molecule in the test set. This is analogous buy ICG-001 to Cramer’s definition: whenever PRESS is larger

than SD, this results in a negative value reflecting complete lack of predictive ability of the training set for the molecules included in the test set (Cramer et al., 1988). Results CoMFA of the β1-adrenoceptor PLS analysis was used in combination with cross-validation to obtain the optimal number of components to be used ubiquitin-Proteasome degradation in the subsequent non-cross-validation analysis. PLS analysis based on least squares fit gave a correlation with a cross-validated \( r^2_\textcv \) of 0.578, with the maximum number of components set equal to five. The non-cross-validated PLS analysis was repeated with the five components, giving an \( r^2_\textncv

\) of 0.993. To obtain statistical confidence limits, the non-cross-validated analysis was repeated with 10 bootstrap groups, which yielded an r 2 of 0.996 (five components,

SEE = 0.027, std dev = 0.003, not steric contribution = 0.558, and electrostatic contribution = 0.442). These parameters are listed in Table 3. The above satisfactory cross-validated correlation coefficient indicates that the CoMFA model is highly reliable. The high bootstrapped r 2 value and low standard deviation suggest a high degree of confidence in the analysis. The calculated biological activities obtained from the analysis are plotted versus the actual values in Fig. 3a. Compounds 9, 10, 11, 15, 18, 23, and 24 (test set) were used to evaluate the predictive power of this CoMFA model. As in the calibration step, a good predictive ability, with an \( r^2_\textpre = 0. 8 4 7 \), for the compounds in the test set was obtained. Table 2 reports that the predicted values fall close to the observed biological activity value, deviating by less than one logarithmic unit. Fig. 3 A graph of experimental vs. predicted activities of the training-set and test-set molecules as β1-AR (a), β2-AR (b), and β3-AR (c) agonists. ( ) Training set; ( ) test set The β1 CoMFA steric and electrostatic fields from the final non-cross-validated analysis are plotted as three-dimensional color contour maps in Figs.