Structural Biochemistry Laboratory, Department of Medicinal Chemistry, Centro de Investigación Príncipe Felipe, Avda. Autopista del Saler 16, 46012 Valencia, Spain.
Bioorg Med Chem. 2010 Oct 1;18(19):7078-84. doi: 10.1016/j.bmc.2010.08.003. Epub 2010 Aug 6.
Solubility plays a very important role in the selection of compounds for drug screening. In this context, a QSAR model was developed for predicting water solubility of drug-like compounds. First, a set of relevant parameters for establishing a drug-like chemical space was defined. The comparison of chemical structures from the FDAMDD and PHYSPROP databases allowed the selection of properties that were more efficient in discriminating drug-like compounds from other chemicals. These filters were later on applied to the PHYSPROP database and 1174 chemicals fulfilling these criteria and with experimental solubility information available at 25°C were retained. Several QSAR solubility models were developed from this set of compounds, and the best one was selected based on the accuracy of correct classifications obtained for randomly chosen training and validation subsets. Further validation of the model was performed with a set of 102 drugs for which experimental solubility data have been recently reported. A good agreement between the predictions and the experimental values confirmed the reliability of the QSAR model.
溶解度在化合物的筛选中起着非常重要的作用。在这方面，开发了一个用于预测类药性化合物水溶性的定量构效关系（QSAR）模型。首先，定义了一组用于建立类药性化学空间的相关参数。通过比较 FDAMDD 和 PHYSPROP 数据库中的化学结构，选择了更有效地将类药性化合物与其他化学物质区分开来的性质。这些筛选标准后来应用于 PHYSPROP 数据库，保留了符合这些标准且在 25°C 下具有实验溶解度信息的 1174 种化合物。从这组化合物中开发了多个 QSAR 溶解度模型，并根据随机选择的训练和验证子集的正确分类准确性选择了最佳模型。该模型还通过一组最近报道了实验溶解度数据的 102 种药物进行了进一步验证。预测值与实验值之间的良好一致性证实了 QSAR 模型的可靠性。
