Quantitative structure-property relationships for predicting metal binding by organic ligands

Quantitative structure-property relationships (QSPRs) are developed to predict the complexation of Al(III), Ca(II), Cd(II), Cu(II), Ni(II), Pb(II), and Zn(II) by organic ligands containing carboxylate, phenol, amine, ether, and alcohol functional groups. These QSPRs predict conditional stability constants (K(M)' at pH 7.0 and l = 0.1) over a range of ligand types with consistent uncertainties of similar to 1 log unit without requiring any steric or connectivity information. Calibration and validation data sets were constructed using 1:1 complex formation constants from the NIST Critical Stability Constants database (version 8.0). The descriptor variables are intuitive quantities conceptually related to metal binding, such as the numbers of various ligand groups, charge density, etc. The resulting calibrations have r(2) = 0.87 to 0.93 and S(pred) = 0.67 to 1.05 log units, with positive values for all ligand count descriptor variables. The GSPRs account for 75-95% of the variability in the validation data set with RMSE of 0.74 to 1.30 log units. These CISPRs improve upon previous work by providing a tested and mechanistically reasonable method for log K(M)' prediction with uncertainties comparable to or better than other CISPRs calibrated with groups of diverse ligands.