Table 1. Top Five δ Receptor Conformationally Sampled Pharmacophore Models That Define the Final Predictive Model^a.

model #	a	X1	b	X2	c	R2	p-value	correlation coefficient
1	0.465	AB	0.514	BN	0.028	0.962	0.00005	0.815
2	0.730	AB	0.235	ANB	–0.031	0.919	0.00054	0.519
3	0.760	AB	0.187	BAN	–0.026	0.911	0.00071	0.442
4	0.749	AB	0.182	ABN	–0.037	0.893	0.00121	0.603
5	1.094	BN	–0.196	ANB	0.108	0.890	0.00133	0.797

^aMultiple regression equations, efficacy = aX₁ + bX₂ + c. N represents the basic nitrogen, A is the aromatic ring and B is the hydrophobic group, as shown in Figure S1 of the Supporting Information. X₁ and X₂ are overlap integrals with respect to the reference compound, while a and b are coefficients for variables X₁ and X₂ and c is the y-intercept in the regression equations. R² is the goodness of fit, p-value is the significance of models, and correlation coefficients between X₁ and X₂ overlap coefficients.