定量结构-性质关系在化合物溶解度预测中的研究进展

Research progress of solubility prediction through quantitative structure-property relationship

溶解度是物质十分重要的一种理化性质,其在化工过程、药物和环境等领域的重要性不可忽视。定量结构-性质关系(quantitative structure-property relationship,QSPR)在化合物溶解度预测中得到广泛的应用。本文介绍了QSPR方法建立溶解度预测模型的研究进展,在总结各类分子描述符和构建溶解度预测模型方法的基础上,分别归纳出三类分子描述符(组成描述符、试验参数及理论计算描述符)和建模方法(线性、非线性及两者联合法),并从不同角度分析它们各自所拥有的特点,比较三类建模的优缺点。最后论述了当前溶解度QSPR研究中存在的不足及未来溶解度预测模型的发展趋势,指出溶解度的预测模型精度有待进一步提高,今后应更关注对化合物在不同p H值、温度、溶剂等更复杂情况下的溶解度预测。

Solubility is one of the major physiochemical properties. The importance of solubility cannot be ignored in chemical processes, drug discovery and environment. This review introduces the latest development of solubility prediction through quantitative structure-property relationship （QSPR） modeling and the ways of molecular description as well. Molecular descriptors are classified as constitutional descriptors, experimental parameters, and theoretical molecular descriptors, and their characteristics are viewed from different aspects. Also, three kinds of modeling methods （linear methods, nonlinear methods, and combination methods） used for solubility prediction are classified and their advantages and disadvantages are summarized. Based on the situation of the present QSPR studies of solubility, new ways are in need to predict solubility in more challenging systems （e.g. solvents at different oH and charged solutes） in the future.