Aqueous prediction software

Large databases on aqueous solubility exist, such as AQUASOL dATAbASE (http //www.pharmacy.arizona.edu/outreach/aquasol/), which contains almost 20,000 solubility records for almost 6,000 compounds, or the already mentioned PhysProp. However, not all situations are covered and the ability to predict this property is still useful. This remark has favoured the development of numerous mathematical models and much prediction software [46]. 

In developing the thermodynamic framework for ECES, we attempted to synthesize computer software that would correctly predict the vapor-liquid-solid equilibria over a wide range of conditions for multicomponent systems. To do this we needed a good basis which would make evident to the user the chemical and ionic equilibria present in aqueous systems. We chose as our cornerstone the law of mass action which simply stated says "The product of the activities of the reaction products, each raised to the power indicated by its numerical coefficient, divided by the product of the activities of the reactants, each raised to a corresponding power, is a constant at a given temperature. ... 

Despite these apparent weaknesses, within the context of a general purpose system for predicting the vapor-liquid-solid equilibria of multicomponent aqueous solutions, GCES as a tool succeeds remarkably well as will be seen in a few illustrations after the following description of the software structure and use. 

Therefore, similar to the attempts made to estimate vapor pressure (Section 4.4) there have been a series of quite promising approaches to derive topological, geometric, and electronic molecular descriptors for prediction of aqueous activity coefficients from chemical structure (e.g., Mitchell and Jurs, 1998 Huibers and Katritzky, 1998). The advantage of such quantitative structure property relationships (QSPRs) is, of course, that they can be applied to any compound for which the structure is known. The disadvantages are that these methods require sophisticated computer software, and that they are not very transparent for the user. Furthermore, at the present stage, it remains to be seen how good the actual predictive capabilities of these QSPRs are. 

Use of commercial software to predict relevant parameters (i.e., aqueous solubility, likely metabolic fate, etc.)... 

Dearden, J.C., Netzeva, T.I., and Bibby, R., Comparison of a number of commercial software programs for the prediction of aqueous solubility, J. Pharm. Pharmacol., 54 (Suppl.) S66, 2002b. 

HazardExpert estimates and takes into account the octanol/water partition coefficient (usually abbreviated to log P or logKow) and pKa of the query compound. The octanol-water partition coefficient is a surrogate for partition between water and fatty biological membranes, which has implications for transport of a chemical to its site of action and its capacity to bind at, or interact with, the site. The pKa, a measure of acidity, indicates the readiness of the compound to ionize in solution, which can have a big influence on partition into membranes, since retention of the ionized form in the aqueous phase is likely to be favored. More recently modules that make use of predictions from ANN have been added to the software package available with HazardExpert. 

Table 4.2 Predictive ability of some commercially available software for aqueous solubility, based on a 122-compound test set of drugs [15].

Solubility in DMSO is less intuitive than aqueous solubility, based on examination of the chemical structure. Chemists can usually differentiate between compounds that are soluble or insoluble in water, but it is much harder to predict compounds that are soluble or insoluble in DMSO (Balakin, 2003). Solubility in DMSO is determined by a subtle balance of oppositely-directed inter- and intramolecular forces. Computational models have been developed to predict DMSO solubility with greater than a 90% success rate (Balakin, 2003 Balakin et al., 2004 Japertas et al., 2004 Lu and Bakken, 2004 Delaney, 2005). Software can be used to provide an alert to a compound with a low DMSO solubility... 

Numerous computational methods have been developed to predict aqueous solubility from molecular structure (Jorgensen and Duffy, 2002 Delaney, 2005). Many of them have an accuracy of 1 log unit. Commercial solubility software is widely available and some of these products estimate the solubility-pH profile for ionizable compounds. Software is most useful for virtual screening of large libraries, prioritization of compounds prior to synthesis and scoring of HTS hits (Oprea et al., 2005). Calculated values can be used to alert teams to potential solubility issues. 

Many drug discovery researchers prefer to use software that can predict aqueous pKa values. For example, the ACD/pk", software program contains a vast library that contains experimental values for over 16,000 compounds in aqueous solutions and over 2000 compounds in nonaqueous solvents [82],... 

In order to better understand some of the nuances associated with the construction and evaluation of predictive models, it is useful to consider actual examples. In this chapter, we will examine a number of datasets containing measured values for aqueous solubility and use these datasets to build and evaluate predictive models. Solubility in water or buffer is an important parameter in drug discovery [13]. Poorly soluble compounds tend to have poor pharmacokinetics and can precipitate or cause other problems in assays. As such, the prediction of aqueous solubility has been an area of high interest in the pharmaceutical industry. Over the last 15 years, numerous papers have been published on methods for predicting aqueous solubility [2, 3, 14]. Although many papers have been published and commercial software for predicting aqueous solubility has been released, reliable solubility prediction remains a challenge. 

An interesting further development of regression trees is the software CUBIST [239], which combines recursive partitioning with linear regression. The predicting functions are RT with LM at terminal nodes. In [40] the aqueous solubility of compounds is modeled using this method. 

Kenny, P.W and Taylor, P.J. (2010 The Prediction of Tautomer Preference in Aqueous Solution, Manuscript Freely available from Openeye Scientific Software or the authors. 

The Prediction of Tautomer Preference in Aqueous Solution, manuscript freely available from OpenEye Scientific Software, pp. 315-320 for Refs. [49] (b)Kenny, P. W. and Taylor, P.J. (2010) The Prediction of Tautomer Preference in Aqueous Solution, manuscript freely available om OpenEye Scientific Software, pp. 320-322 for Ref. [38] (c)Kenny, P. W. and Taylor, P.J. (2010) The Prediction of Tautomer Preference in Aqueous Solution, manuscript freely available from OpenEye Scientific Software, pp. 323-326 for Refs. [50] and [51]. 

Prediction of Tautomer Preference in Aqueous Solution, Openeye Scientific Software. 

---