Department of Chemistry, Georgia State University, Atlanta, GA 30303, USA.
Bioorg Med Chem. 2011 Aug 1;19(15):4552-61. doi: 10.1016/j.bmc.2011.06.026. Epub 2011 Jun 16.
A dataset of 55 compounds with inhibitory activity against Leishmania donovani axenic amastigotes and Leishmania amazonensis intracellular parasites was examined through three-dimensional quantitative structure-activity relationship modeling employing molecular descriptors from both rigid and flexible compound alignments. For training and testing purposes, the compounds were divided into two datasets of 45 and 10 compounds, respectively. Statistically significant models were constructed and validated via the internal and external predictions. For all models employing steric, electrostatic, hydrophobic, H-donor and H-acceptor molecular descriptors, the R² values were greater than 0.90 and the SEE values were less than 0.22. The models obtained from rigid and flexible compounds were employed together to obtain a conservative method for predictions. This method minimized under predictions. Molecular descriptors from the models were then extrapolated, for the overall predictive devices and the individual compounds, and examined with regard to inhibitory activity. Information gained from the molecular descriptors is useful in the design of novel compounds. The models obtained can be employed to predict activities of the compounds designed and/or form predictions for compounds that exist and have not yet been examined with biological inhibitory assays.
研究了一个包含 55 种化合物的数据集，这些化合物对利什曼原虫无鞭毛体和利什曼原虫内寄生虫具有抑制活性，通过使用刚性和柔性化合物排列的分子描述符进行三维定量构效关系建模进行了检查。为了培训和测试目的，将化合物分别分为 45 种和 10 种化合物的两个数据集。通过内部和外部预测构建和验证了具有统计学意义的模型。对于所有使用立体、静电、疏水、H 供体和 H 受体分子描述符的模型，R²值均大于 0.90，SEE 值均小于 0.22。将来自刚性和柔性化合物的模型一起用于获得一种保守的预测方法。该方法最小化了预测不足。然后对整体预测设备和个别化合物进行了从模型中得出的分子描述符的外推，并根据抑制活性对其进行了检查。从分子描述符中获得的信息可用于设计新型化合物。获得的模型可用于预测设计化合物的活性和/或对存在但尚未用生物抑制测定法进行检查的化合物进行预测。
