Aqueous Drug Solubility: What Do We Measure, Calculate and QSPR Predict?

Abstract

Detailed critical analysis of publications devoted to QSPR of aqueous solubility is presented in the review with discussion of four types of aqueous solubility (three different thermodynamic solubilities with unknown solute structure, intrinsic solubility, solubility in physiological media at pH=7.4 and kinetic solubility), variety of molecular descriptors (from topological to quantum chemical), traditional statistical and machine learning methods as well as original QSPR models.

ADME Prediction with KNIME: In silico aqueous solubility models based on supervised recursive machine learning approaches

“…RMSE = [ 1/n Σi (yi obs -yi calc ) 2 ] 1/2 , where y obs / y calc = observed/calculated value of log S according to model, n = number of measurements of log S. r 2 : the squared of correlation coefficient of regression. r 2 = 1 -Σi (yi obs -yi calc ) 2 / Σi (yi obs -<y>) 2 , where y=log S, and <y> is the mean value of log S.…”

“…Aqueous solubility is one of the most important physicochemical properties determined during the drug discovery and development processes [ 1 ]. It is considered a relevant parameter during ADME (Absorption, Distribution, Metabolism and Excretion) studies [ 2 ], and it is a key factor that can affect oral absorption and bioavailability of drugs [ 3 ]. Considering that approximately 40 % of drugs on the market and about 75 % of compounds in development have a poor aqueous solubility, early identification of this property should reduce failures in the pharmaceutical development process [ 4 ].…”

“…However, the determination of experimental solubility proves to be difficult, time-consuming and too expensive, or unrealistic to test thousands or millions of compounds used in high throughput screening (HTS) [ 6 ]. In this respect, the in-silico prediction of aqueous solubility by Quantitative Structure-Property Relationship (QSPR) has been widely used in the early stage of the drug discovery and development process [ 2 ]. Several QSPR models for predicting aqueous solubility have been developed in recent decades, but the performance of some of these models on various solubility datasets has demonstrated the poor reliability of the prediction methods [ 2 , 5 , 7 , 8 ].…”

Pushing the limits of solubility prediction via quality-oriented data selection

“…There are several publicly available resources to calculate molecular descriptors (Yap 2011;. Most 2D descriptors are calculated with absolute accuracy while the 3D descriptors carry the errors of the methodological approximations they have been calculated with (Raevsky et al, 2019). Admitting that the 3D descriptors provide more detailed information, such as atomic distances and energy data of the compounds, there is yet no clear evidence about their impacts on the solubility predictions (Balakin et al 2006;Gao et al, 2020;Yan et al, 2004;Salahinejad et al, 2013).…”

Pushing the limits of solubility prediction via quality-oriented data selection

“…There are several publicly available resources to calculate molecular descriptors [21,22,23]. Most 2D descriptors are calculated with absolute accuracy while the 3D descriptors carry the errors of the methodological approximations they have been calculated with [24]. Admitting that the 3D descriptors provide more detailed information, such as atomic distances and energy data of the compounds, there is yet no clear evidence about their impacts on the solubility predictions [8,25,26,27].…”