Oltulu Oral, Yaşar Mehmet M, Eroğlu Erol
Harran University, Physics Department, Osmanbey Kampus, 6300 Sanliurfa, Turkey.
Eur J Med Chem. 2009 Sep;44(9):3439-44. doi: 10.1016/j.ejmech.2009.02.016. Epub 2009 Feb 21.
In this study, we present an application of EVA descriptors for a QSAR model of inhibition of carbonic anhydrase isozyme CA II by an heterogeneous set of 66 sulfonamide compounds. For each of the compounds, geometry optimization and frequency calculations have been performed using the DFT/B3LYP level of the theory in conjugated with the 6-31G* basis set. Different numbers of EVA descriptors for each structure were produced by applying various values of Gaussian kernel of a fixed standard deviation, sigma (cm(-1)) and sampled at fixed increments of L (cm(-1)) during the evaluation of the descriptors based on their vibrational frequencies. The set of compounds was divided into two subsets. The first subset contained the 22 compounds that were used as the test compounds. The remaining 44 compounds were used as the training set. Several QSAR models have been developed using these calculated EVA descriptors and the carbonic anhydrase isozyme CA II inhibitory data (K(i)) of the compounds. Among the QSAR models evaluated, the one that produced the best statistical results had the parameters sigma and L both equal to 5 cm(-1). This model produced correlation values (R(2)) of 0.777 and 0.616 for the training and test sets, respectively. The results of this study showed that EVA descriptors perform well as explanatory and predictive tools for modeling the inhibition activity of carbonic anhydrase by a set of sulfonamide compounds.
在本研究中，我们展示了将EVA描述符应用于由66种磺酰胺化合物组成的异构集对碳酸酐酶同工酶CA II抑制作用的QSAR模型。对于每种化合物，使用密度泛函理论（DFT）/B3LYP水平结合6-31G*基组进行了几何优化和频率计算。通过应用固定标准偏差sigma（cm⁻¹）的高斯核的不同值，并在基于其振动频率评估描述符期间以固定的L（cm⁻¹）增量进行采样，为每个结构生成了不同数量的EVA描述符。该化合物集被分为两个子集。第一个子集包含用作测试化合物的22种化合物。其余44种化合物用作训练集。使用这些计算出的EVA描述符和化合物的碳酸酐酶同工酶CA II抑制数据（K(i)）开发了几个QSAR模型。在评估的QSAR模型中，产生最佳统计结果的模型的参数sigma和L均等于5 cm⁻¹。该模型对训练集和测试集的相关值（R²）分别为0.777和0.616。本研究结果表明，EVA描述符作为一组磺酰胺化合物对碳酸酐酶抑制活性建模的解释和预测工具表现良好。
