非负矩阵因子分解算法解析手性药物重叠峰的HPLC-DAD数据

Resolution of HPLC-DAD data of chiral drugs with overlapping peaks by non-negative matrix factorization

目的:研究非负矩阵因子分解算法(NMF)用于手性药物HPLC-DAD二维数据解析的可行性及其影响因素。方法:根据化学波谱的基本特征如色谱的单峰性改进NMF算法。考查组分之间色谱分离度、光谱相似程度以及迭代次数对模拟重叠峰解析结果的影响,并将NMF算法应用于实测体系盐酸舍曲林对映异构体体(cis-1R,4R;cis-1S,4S)重叠峰解析。结果:解析结果表明,在色谱严重重叠,光谱相似甚至完全相同时,在合适的迭代次数下,NMF能解析出重叠峰中各单组分的光谱和相应色谱。结论该种二维数据的解析方法,将为混合样本特别是手性药物混合体系的分析提供新途径。

Aim：To study the feasibility and influential factors for the resolution of HPLC-DAD data of chiral drugs by non-negative matrix factorization （NMF） algorithm. Methods： NMF algorithm was employd according to the unimodality of chromatograms of chemical signals. The simulated two-dimensional （2-D） data was resolved by the modified NMF thereby investigate the influence of iterative number, chromatographic resolution and spectral similarity on the resolution of overlapping peaks. Real HPLC-DAD data of sertraline hydrochloride enantiomers（ cis-1 R, 4R c/s-1S, 4S ） was resolved by the NMF. Results： The NMF was capable of identification of both the spectra and chromatograms from the overlapping peaks respectively with the appropriately iterative number, especially for the similar or identical ultraviolet spectra and badly overlapping chromatograms. Conclusion： The NMF is a promising chemometric resolution method for complex samples, especially for chiral drugs.