Global Spectral Deconvolution Based on Non-Negative Matrix Factorization in GC × GC–HRTOFMS

• Published in: Analytical chemistry (Washington) Vol. 87; no. 3; pp. 1829 - 1838
• Main Authors: Zushi, Yasuyuki, Hashimoto, Shunji, Tanabe, Kiyoshi
• Format: Journal Article
• Language: English
• Published: United States American Chemical Society 03.02.2015
• Subjects: Algorithms; Databases, Factual; Gas Chromatography-Mass Spectrometry - methods; Workflow
• ISSN: 0003-2700; 1520-6882
• DOI: 10.1021/ac5038544

A global spectral deconvolution, based on non-negative matrix factorization (NMF) in comprehensive two-dimensional gas chromatography high-resolution time-of-flight mass spectrometry, was developed. We evaluated the ability of various instrumental parameters and NMF settings to derive high-performance detection in nontarget screening using a sediment sample. To evaluate the performance of the process, a NIST library search was used to identify the deconvoluted spectra. Differences of the instrumental scan rates (25 and 50 Hz) in deconvolution were evaluated and results show that a high scan rate enhanced the number of compounds detected in the sediment sample. A higher mass resolution in the range of 1 000 to 10 000 and a higher m/z precision in the deconvolution were needed to obtain an accurate mass database. After removal of multiple duplicate hits, which occurred in batch processes of NIST library search on the deconvolution result, 62 unique assignable spectra with a match factor ≥900 were obtained in the deconvoluted chromatogram from the sediment sample, including 54 spectra that were refined by the deconvolution. This method will help to detect and build up well-resolved reference spectra from various complex mixtures and will accelerate nontarget screening.