Komsta Łukasz
Department of Medicinal Chemistry, Medical University of Lublin, Jaczewskiego 4, 20-090 Lublin, Poland.
Anal Chim Acta. 2008 Nov 23;629(1-2):66-72. doi: 10.1016/j.aca.2008.09.036. Epub 2008 Sep 19.
A QSRR approach, based on functional group counts, useful for prediction of the retention of any small organic solute with small molecule in seven TLC screening systems is presented. The equations are based on 154 possible substituents as the only molecular descriptors involved. The model optimization was performed by uninformative variable elimination-partial least squares (UVE-PLS), reducing the variables involved. The final equations, with 7-32 variables, explain 32-60% of overall explained variance and their predictive ability (Q(2)) includes in range 25-54% (LOO crossvalidation) and 12-43% (external validation). Their simplicity allows to computation without any complicated software.
本文提出了一种基于官能团计数的QSRR方法，该方法可用于预测七种薄层色谱筛选系统中任何含小分子的有机溶质的保留情况。这些方程仅基于154种可能的取代基作为唯一涉及的分子描述符。通过无信息变量消除-偏最小二乘法（UVE-PLS）进行模型优化，减少了所涉及的变量。最终方程包含7至32个变量，解释了32%至60%的总解释方差，其预测能力（Q(2)）在25%至54%（留一法交叉验证）和12%至43%（外部验证）范围内。其简单性使得无需任何复杂软件即可进行计算。
