化学
加合物
数量结构-活动关系
可转让性
质子化
流出物
质谱法
色谱法
分析化学(期刊)
响应系数
质量
离子
质谱
有机化学
立体化学
统计
环境工程
工程类
罗伊特
数学
作者
Selina Tisler,Kristoffer Kilpinen,David I. Pattison,Giorgio Tomasi,Jan H. Christensen
标识
DOI:10.1021/acs.analchem.3c03791
摘要
Quantitative nontarget analysis (qNTA) for liquid chromatography coupled to high-resolution mass spectrometry enables a more comprehensive assessment of environmental samples. Previous studies have shown that correlations between a compound's ionization efficiency and a range of molecular descriptors can predict the compound's concentration within a factor of 5. In this study, the qNTA approach was further improved by considering all mass adducts instead of only the protonated ion. The model was based on a quantitative structure-property relationship (QSPR), including 216 contaminants of emerging concern (CECs), of which 80 exhibited adduct formation that accounted for >10% of the total peak intensity. When all mass adducts were included, the test set coefficient of determination improved to Q2 = 0.855 compared to Q2 = 0.670 when only the protonated ions were considered (test set median RF error factor 1.6). The inclusion of all adducts was also important to transfer the RF QSPR model reliably. It was assumed that RF variations are sequence-dependent; therefore, a second QSPR model for the prediction of the transferability factor was built for each sequence. For validation, samples were analyzed up to two years apart. The median prediction fold change was 1.74 for analytical standards (63 compounds) and 2.4 for enriched wastewater effluent samples (41 compounds), with 80% of the compounds predicted within a fold change of 2.4 and 3.3, respectively. The model was also validated on a second instrument, where 80% of the 26 compounds in wastewater effluent were predicted within a factor of 3.8.
科研通智能强力驱动
Strongly Powered by AbleSci AI