engleski

Modeling the octanol-water partition coefficient by a modified molecular connectivity index

The logarithm of the partition coefficient between n-octanol and water, log P, is an important physico-chemical parameter which frequently appears in QSAR (quantitative structure-activity relationship) models. Because of difficulties in the experimental determination of this parameter and the impossibility of its determination for unsynthesized compounds, a number of methods have been proposed for calculating log P. The first-order molecular connectivity index, one of the most popular graph-theoretical molecular descriptors, has been successfully used to predict lipophilicity in sets of closely related compounds, but is less suitable for groups of heterogeneous compounds. To circumvent this obvious shortcoming, here we propose a modified method for the calculation of the molecular connectivity index, which is essentially based on the optimization of delta values of corresponding skeletal atoms. We applied the thus modified index to generate a common model describing the lipophilicity of several structurally diverse groups of compounds (alkanes, alkenes, alkynes, cycloalkanes, aliphatic alcohols, ethers, esters, ketones, carboxylic acids, esters, amines, amides, halogenated alkanes, and some benzene derivatives) and found statistically significant improvement compared to the model based on the standard molecular connectivity index. We also tested the predictive ability of the common model and found it satisfactory. Preliminary data suggest that additional correction factors are needed for the estimation of log P values of more complex compounds, in a similar fashion as, for instance, in the Hansch-Leo and Rekker methods. It appears that the proposed optimization method could provide interesting clues towards a coherent model for the estimation of log P in the framework of the molecular connectivity formalism.

n-octanol-water partition coefficient; molecular connectivity index

nije evidentirano