Table 4.

Equations used to predict logKeq values for DNA-binding drugs^a.

	Predictive equation	r	p	AdR2
All drugs	log Keq = -0.255(± 0.100) XlogP - 0.003(± 0.002)PSA + 6.603(± 0.465)	0.603	6.6 × 10-2	0.258
All drugs	log Keq = -0.181(± 0.097)XlogP + 5.865(± 0.215)	0.461	8.4 × 10-2	0.152
Intercalators	log Keq = -0.225(± 0.090)XlogP + 6.054(± 0.229)	0.662	3.7 × 10-2	0.368
'M-region'	log Keq = 0.128(± 0.067) XlogP - 0.178(± 0.020)HbA + 0.173(± 0.066)HbD + 7.577(± 0.268)	0.984	2.0 × 10-3	0.944
'M-region'	log Keq = -0.138(± 0.190)HbA + 8.090(± 0.303)	0.955	2.2 × 10-4	0.897

The predictive equations are presented for the three sets of drugs analyzed (All DNA-binding drugs, Intercalators and 'M-region' compounds) described in the main text^b.

^aObtained by multiple regression analysis, in which molecular descriptors showing multicollinearity were discarded (see the main text for details). The predictive equations displayed are those statistically "more significant" for each set of predictors (actual p values, ANOVA test, are shown in the Table), r is the correlation coefficient of the linear fit, AdR² is the fraction of the variance in logKeq that is explained (predicted) by the model, corrected for the number the variables in the model, as described in Methods.

^bThe cases (drugs) used in the calculations for each set were 15, 10 and 8 respectively.