Liu Huitao, Wen Yingying, Luan Feng, Gao Yuan, Li Xiuyong
Department of Applied Chemistry, Yantai University, No. 32 Qingquan Road, Yantai 264005, PR China.
Anal Chim Acta. 2009 Sep 1;649(1):52-61. doi: 10.1016/j.aca.2009.07.013. Epub 2009 Jul 10.
QSPR models for the prediction of UV maximum absorption wavelength (lambda(max)) of 69 flavones were developed based on their structures alone. A six-descriptor linear correlation by heuristic method (HM) and a nonlinear model using radial basis function neural network (RBFNN) approach were reported. The statistical parameters provided by the HM model (R2=0.961, F=207.820, RMS=6.555 for the training set and R2=0.967, F=293.218, RMS=7.176 for the test set) and the RBFNN model (R2=0.971, F=1826.086, RMS=5.350 for the training set, and R2=0.978, F=452.512, RMS=5.722 for the test set) indicated satisfactory stability and predictive ability. The descriptors appearing in these models are discussed. This QSPR approach is suitable for the prediction of maximum absorption wavelength of flavones, and can contribute to a better understanding of structural factors of the organic compounds responsible for it.
仅基于69种黄酮类化合物的结构开发了用于预测其紫外最大吸收波长（λ(max)）的定量构效关系（QSPR）模型。报道了一种通过启发式方法（HM）建立的六描述符线性相关性模型以及一种使用径向基函数神经网络（RBFNN）方法的非线性模型。HM模型提供的统计参数（训练集：R2 = 0.961，F = 207.820，RMS = 6.555；测试集：R2 = 0.967，F = 293.218，RMS = 7.176）和RBFNN模型提供的统计参数（训练集：R2 = 0.971，F = 1826.086，RMS = 5.350；测试集：R2 = 0.978，F = 452.512，RMS = 5.722）表明模型具有令人满意的稳定性和预测能力。讨论了这些模型中出现的描述符。这种QSPR方法适用于黄酮类化合物最大吸收波长的预测，有助于更好地理解导致该现象的有机化合物的结构因素。
