Abstract:A novel method for the estimation of aqueous solubility was solely based on simple atom contribution. Each atom in a molecule has its own contribution to aqueous solubility and was developed. Altogether 76 atom types were used to classify atoms with different chemical environments. Moreover, two correction factors, including hydrophobic carbon and square of molecular weight, were used to account for the inter-/intramolecular hydrophobic interactions and bulkiness effect. The contribution coefficients of different atom types and correction factors were generated based on a multiple linear regression using a learning set consisting of 1290 organic compounds. The obtained linear regression model possesses good statistical significance with an overall correlation coefficient (r) of 0.96, a standard deviation (s) of 0.61, and an unsigned mean error (UME) of 0.48. The actual prediction potential of the model was validated through an external test set with 21 pharmaceutically and environmentally interesting compounds. For the test set, a predictive r=0.94, s=0.84, and UME=0.52 were achieved. Comparisons among eight procedures of solubility calculation for those 21 molecules demonstrate that our model bears very good accuracy and is comparable to or even better than most reported techniques based on molecular descriptors. Moreover, we compared the performance of our model to a test set of 120 molecules with a popular group contribution method developed by Klopman et al. For this test set, our model gives a very effective prediction (r=0.96, s=0.79, UME=0.57), which is obviously superior to the predicted results (r=0.96, s=0.84, UME=0.70) given by the Klopman's group contribution approach. Because of the adoption of atoms as the basic units, our addition model does not contain a "missing fragment" problem and thus may be more simple and universal than the group contribution models and can give predictions for any organic molecules. A program, drug-LOGS, had been developed to identify the occurrence of atom types and estimate the aqueous solubility of a molecule.

A New Atom-Additive Method for Calculating Partition Coefficients

Calculating partition coefficient by atom-additive method

A New Method for Calculating Partition Coefficients of Organic Compounds

Computation of octanol-water partition coefficients by guiding an additive model with knowledge.

A New Model for Calculating Logp for Organic Compounds

Calculating Partition Coefficients of Peptides by the Addition Method

An Empirical Additive Model for Aqueous Solubility Computation: Success and Limitations

Computational Approaches to Estimating Octanol-Water Partition Coefficient and Water Solubility of Organic Compounds

Calculating the Partition Coefficients of Organic Solvents in Octanol/Water and Octanol/Air

The Octanol-Water Partition Coefficient: The Effect of Electronic Interactions

The Octanol-Water Partition Coefficient:The Effect of Electronic Interactions

ADME Evaluation in Drug Discovery. 4. Prediction of Aqueous Solubility Based on Atom Contribution Approach

Dimensionally reduced machine learning model for predicting single component octanol-water partition coefficients

Comparison of Four Methods of Predicting Newly Measured Octanol/water Coefficients (log Kow) for Heterocyclic Nitrogen Compounds and the Partition Mechanism

Calculation of Peptide'S Partition Coefficients by Amino Acid Addition Model

Effect of Atomic Charges on Octanol–Water Partition Coefficient Using Alchemical Free Energy Calculation

ADME Evaluation in Drug Discovery. Part 2. Prediction of Partition Coefficient by Atom‐Additive Approach Based on Atom‐Weighted Solvent Accessible Surface Areas.

Collection of Partition Coefficients in Hexadecyltrimethylammonium Bromide, Sodium Cholate, and Lithium Perfluorooctanesulfonate Micellar Solutions: Experimental Determination and Computational Predictions

Molecular Partition Coefficient from Machine Learning with Polarization and Entropy Embedded Atom- Centered Symmetry Functions

OWPCP: A Deep Learning Model to Predict Octanol-Water Partition Coefficient

ADME Evaluation in Drug Discovery. 2. Prediction of Partition Coefficient by Atom-Additive Approach Based on Atom-Weighted Solvent Accessible Surface Areas [J. Chem. Inf. Comput. Sci. 43, 1058−1067 (2003)]