The Prediction of Water Solubility and of pK_a-Values by Physicochemical Descriptors
Lecture of Prof. J. Gasteiger

The Prediction of Water Solubility and of
pK_a-Values by Physicochemical Descriptors

Outline

Prediction of Solubility
Objectives

Prediction of Solubility
Procedure

Dataset of Water Solubility of
Organic Compounds

Modeling Water Solubility of
Organic Compounds

Physicochemical Descriptors Calculated by PETRA

Classification of Water Solubility
of Organic Compounds

Modeling Water Solubility of
Organic Compounds

3D Structure Coding: Radial Distribution Function

Radial Distribution Function: RDF Code

Representation of Structures

Training / Test Data Set
by Kohonen Neural Network

Modeling of Solubility: Training of Backpropagation Neural Network

Modeling of Solubility: Prediction of Test Set

Modeling of Solubility by a Backpropagation
Neural Network

Datasets

Modeling of Solubility by a Backpropagation
Neural Network

Comparision of datasets

Prediction of Solubility
Conclusions

Prediction of pK_a-Values

Prediction of pK_a-Values

Fundamental Effects on Acidity

A Simple Empirical Model for Deprotonation

pK_a-Values of O, N, and S Compounds: Data

pK_a-Values of Oxygen Compounds

pK_a-Values of Oxygen Compounds

pK_a-Values of Oxygen Compounds

Descriptors from Quantum Mechanical Calculations

pK_a-Values of Oxygen Compounds

Comparison of Empirical Linear Models for
pK_a-Values of O, N, and S Compounds

Comparison of Empirical Linear Models for
pK_a-Values of O, N, and S Compounds

Prediction of pK_a-Values

Further Work

Acknowledgements