论文部分内容阅读
本文利用多变量法及偏最小二乘法（PLS）将化合物化学性质的定量信息与柱子和流动相的定量指标结合起来，建成可以预测相关的新化合物保留时间的回归模型。该程序用核苷类似物的数据进行验证，以7种化合物的多达28种柱子与流动相组合（每个化合物3～5种）的色谱研究数据建立PLS模型，以此模型预测另9种物质的保留时间，并与实测数据相比较。预测值为实测值的115±82％（95％置信限）。
In this paper, we use the multivariate and partial least squares (PLS) methods to combine the quantitative information of the chemical properties of the compounds with the quantitative indicators of the mobile phase and the mobile phase to establish a regression model that predicts the relative retention times of the new compounds. The program was validated using data from nucleoside analogs and PLS models were established using chromatographic data from up to 28 columns of 7 compounds combined with a mobile phase (3-5 compounds per compound) to predict the other 9 Substance retention time, and compared with the measured data. The predicted value is 115 ± 82% of the measured value (95% confidence limit).