Prediction of aqueous solubility of organic compounds by the general solubility equation (GSE)

The revised general solubility equation (GSE) is used along with four different methods including Huuskonen's artificial neural network (ANN) and three multiple linear regression (MLR) methods to estimate the aqueous solubility of a test set of the 21 pharmaceutically and environmentally interesting compounds. For the selected test sets, it is clear that the GSE and ANN predictions are more accurate than MLR methods. The GSE has the advantages of being simple and thermodynamically sound. The only two inputs used in the GSE are the Celsius melting point (MP) and the octanol water partition coefficient (K(ow)). No fitted parameters and no training data are used in the GSE, whereas other methods utilize a large number of parameters and require a training set. The GSE is also applied to a test set of 413 organic nonelectrolytes that were studied by Huuskonen. Although the GSE uses only two parameters and no training set, its average absolute errors is only 0.1 log units larger than that of the ANN, which requires many parameters and a large training set. The average absolute error AAE is 0.54 log units using the GSE and 0.43 log units using Huuskonen's ANN modeling. This study provides evidence for the GSE being a convenient and reliable method to predict aqueous solubilities of organic compounds.     

Estimation of aqueous solubility of organic molecules by the group contribution approach. Application to the study of biodegradation.

A reliable and generally applicable aqueous solubility estimation method for organic compounds based on a group contribution approach has been developed. Two models have been established based on two different sets of parameters. One has a higher accuracy, while the other has a more general applicability. The prediction potentials of these two models have been evaluated through cross-validation experiments. For model I, the mean cross-validated r2 and SD for 10 such cross-validation experiments were 0.946 and 0.503 log units, respectively. While for model II, they were 0.953 and 0.546 log units, respectively. Applying our models to estimate the water solubility values for the compounds in an independent test set, we found that model I can be applied to 13 out of 21 compounds with a SD equal to 0.58 log unit and model II can be applied to all the 21 compounds with a SD equal to 1.25 log units. Our models compare favorably to all the current available water estimation methods. A program based on this approach has been written in FORTRAN77 and is currently running on a VAX/VMS system. The program can be applied to estimate the water solubility of the water solubility of any organic chemical with a good or fairly good accuracy except for except for electrolytes. Applying our aqueous solubility estimation models to biodegradation studies, we found that although the water solubility was not the sole factor controlling the rate of biodegradation, ring compounds with greater solubilities were more likely to biodegrade at a faster rate. The significance of the relationship between water solubility and biodegradation activity has been illustrated by predicting the biodegradation activity of 27 new chemicals based solely on their estimated solubility values.     

Indirect photolysis of organic compounds: prediction of OH reaction rate constants through molecular orbital calculations

The MOOH approach is a perturbational molecular orbital method to predict rate constants of indirect photolysis of organic compounds through reaction with OH radicals. It employs the semiempirical AM1 scheme as the underlying quantum chemical model. The original method introduced by Klamt has been reparametrized using an up-to-date set of 675 compounds with experimental rate constants and outperforms the prominent Atkinson increment scheme for this training set as well as for an extended set of 805 compounds, yielding an overall root-mean-square error of 0.32 log units. The discussion includes detailed comparative analyses of the model performances for individual compound classes. The present model calibration refers mainly to monofunctional compounds but performs already reasonably well for multifunctional compounds. For predictive applications, both the Atkinson scheme and the alternative, independent AM1-MOOH model can be used as components of a consensus modeling approach, arriving at increased confidence in cases where the different models agree.     

Prediction of Chromatographic Retention of Pyrazine and Alkylpyrazines in RP-LC

Retention prediction models for a group of pyrazines chromatographed under reversed-phase mode were developed using multiple linear regression (MLR) and artificial neural networks (ANNs). Using MLR, the retention of the analytes were satisfactorily described by a two-predictor model based on the logarithm of the partition coefficient of the analytes (log P) and the percentage of the organic modifier in the mobile phase (ACN or MeOH). ANN prediction models were also derived using the predictors derived from MLR as inputs and log k as outputs. The best network architecture was found to be 2-2-1 for both ACN and MeOH data sets. The optimized ANNs showed better predictive properties than the MLR models especially for the ACN data set. In the case of the MeOH data set, the MLR and ANN models have comparable predictive performance.     

Prediction of pH-dependent aqueous solubility of druglike molecules

In the present work, the Henderson-Hasselbalch (HH) equation has been employed for the development of a tool for the prediction of pH-dependent aqueous solubility of drugs and drug candidates. A new prediction method for the intrinsic solubility was developed, based on artificial neural networks that have been trained on a druglike PHYSPROP subset of 4548 compounds. For the prediction of acid/base dissociation coefficients, the commercial tool Marvin has been used, following validation on a data set of 467 molecules from the PHYSPROP database. The best performing network for intrinsic solubility predictions has a cross-validated root mean square error (RMSE) of 0.70 log S-units, while the Marvin pKa plug-in has an RMSE of 0.71 pH-units. A data set of 27 drugs with experimentally determined pH-solubility curves was assembled from the literature for the validation of the combined pH-dependent model, giving a mean RMSE of 0.79 log S-units. Finally, the combined model has been applied on profiling the solubility space at low pH of five large vendor libraries.     

Multilinear regression and comparative molecular field analysis (CoMFA) of azo dye-fiber affinities. 2. Inclusion of solution-phase molecular orbital descriptors

For a data set with 30 direct azo dyes taken from literature, quantitative structure-activity relationship (QSAR) analyses have been performed to model the affinity of the dye molecules for the cellulose fiber. The electronic structure of the compounds was characterized using quantum chemical gas-phase (AM1) and continuum-solvation molecular orbital parameters. As regards the solution phase, COSMO appears to be better suited than SM2 in quantifying relative trends of the aqueous solvation energy. For the dye-fiber affinity, the leave-one-out prediction capability of multilinear regression equations is superior to CoMFA, with predictive squared correlation coefficients ranging from 0.63 (pure CoMFA) to 0.89. At the same time, solution-phase CoMFA is superior to previously derived AM1-based CoMFA models. As a general trend, the dye-fiber affinity increases with increasing electron donor capacity that corresponds to an increasing hydrogen bond acceptor strength of the azo dyes. The discussion includes the consideration of structural features that are likely to be involved in dye-fiber and dye-dye hydrogen bonding interactions, and possible links between CoMFA electrostatic results and the atomic charge distribution of the compounds.