Deep understanding of structure–solubility relationship for a diverse set of organic compounds using matched molecular pairs

Aqueous solubility is an important biopharmaceutical property in drug discovery and development. Although it has been studied for decades, the impact on solubility by the substructures (or fragments) of compounds are still not fully understood and characterized. This study aims to obtain fragment–solubility relationships using matched molecular pairs, and to provide further insight and suggestions for chemists on structural modifications to improve solubility profiles of drug-like molecules. A set of 2794 compounds with measured intrinsic aqueous solubility (log S ) was fragmented into rings, linkers, and R groups using a controlled hierarchical fragmentation method. Then matched molecular pairs that differ by only one chemical transformation (i.e., addition or substitution of fragments) were identified and analyzed. The difference in solubility for each matched molecular pair was calculated, and the impact of the corresponding chemical transformation on solubility was investigated. The final product of this study was a fragment–solubility knowledgebase containing relative contributions to solubility of various medicinal chemistry design elements (R-groups, linkers, and rings). Structural modifications that might improve solubility profiles, that is, addition/deletion/substitution of fragments, could be derived from this knowledgebase. This knowledgebase could be used as an expert tool in lead optimization to improve solubility profiles of compounds, and the analysis method could be applied to study other biological and ADMET properties of organic compounds.