Abstract
In a recent note to this journal Moriguchi et al. compared the reliability of several methods for calculating log P octanol/water from structure. The structures chosen were 22 drugs which Rekker and Manhold had analyzed earlier. On the basis of these values a statistical analysis indicated that the reliability of the methods decreased in the order : Moriguchi, Hansch, Rekker and Suzuki. The fact that regression equations relating measured to calculated values produced such unreasonably high deviations (s=0.764 to 1.240) and poor correlation coefficients (0.701 to 0.901) should have alerted the authors that some of the measured log P value were in error. Three major errors are apparent, and these affect the statistics to such an extent that the comparisons are not valid.
