2020-06-04, Wyttenbach, Nicole, Niederquell, Andreas, Kuentz, Martin

There has been much recent interest in machine learning (ML) and molecular quantitative structure property relationships (QSPR). The present research evaluated modern ML-based methods implemented in commercial software (COSMOquick and Molecular Modeling Pro), compared to a classical group contribution approach (Joback and Reid method), to estimate melting points and enthalpy of fusion values. A broad data set of market compounds was gathered from the literature, together with new data measured by differential scanning calorimetry for drug candidates. The highest prediction accuracy was achieved by QSPR using stochastic gradient boosting. The model deviations were discussed, particularly the implications on thermodynamic solubility modeling, as this typically requires estimation of both melting point and enthalpy of fusion. The results suggested that despite considerable advancement in prediction accuracy, there are still limitations especially with complex drug candidates. It is recommended that in such cases, melting properties obtained in silico should be used carefully as input data for thermodynamic solubility modeling. Future research will show how the prediction limits of thermophysical drug properties can be further advanced by even larger data sets and other ML algorithms or also by using molecular simulations.

2018-07, Niederquell, Andreas, Wyttenbach, Nicole, Kuentz, Martin

Solubility parameters have been applied extensively in the chemical and pharmaceutical sciences. Particularly attractive is calculation of solubility parameters based on chemical structure and recently, new in silico methods have been proposed. Thus, screening charge densities of molecular surfaces (i.e. so-called σ-profiles) are used by the conductor-like screening model for real solvents (COSMO-RS) and can be employed in a quantitative structure property relationship (QSPR) to predict solubility parameters. In the current study, it was aimed to compare both in silico methods with an experimental dataset of pharmaceutical compounds, which was complemented with own measurements by inverse gas chromatography. An initial evaluation of the total solubility parameters of reference solvents resulted in excellent predictions (observed versus predicted values) with R2 of 0.855 (COSMO-RS) and 0.945 (QSPR). The subsequent main study of pharmaceutical compounds exhibited R2 values of 0.701 (COSMO-RS) and 0.717 (QSPR). The comparatively lower prediction was to some extent due to the solid state of pharmaceuticals with known conceptual limitations of the solubility parameter and possible experimental bias. Total solubility parameters were also estimated by classical group contribution methods, which had comparatively lower prediction power. Therefore, the new in silico methods are highly promising for pharmaceutical applications.