Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS) is the state-of-the-art technique for nontargeted elucidation of the molecular composition of dissolved organic matter (DOM). Peak reproducibility is a major factor that affects the reliability of FT-ICR MS spectra due to challenges involved with the efficient discrimination of noise peaks from thousands of true positive peaks in DOM. A noise removal algorithm, outlier spectra identification algorithm, and occurrence filter have been incorporated into a newly proposed workflow, namely parallel samples representation by similar samples (PSRSS), to address the above challenges. The performance of the workflow was validated by instrumental and experimental replicates. Inspection of isotopic and homologous peaks in the PSRSS workflow enabled elimination of noise peaks, significantly decreased the Bray-Curtis dissimilarity (from 54.83 ± 2.62% and 60.05 ± 6.36% to 33.21 ± 1.14% and 41.46% ± 6.06% for instrumental and experimental replicates and from 66.14 ± 11.50% and 64.50 ± 2.38% to 53.91 ± 15.58% and 50.15 ± 3.65% for independent samples), and improved the molecular formula assignment ratio (from 42.69 ± 12.87% to 58.22 ± 12.58%  and 45.50 ± 1.89% to 60.79 ± 2.65% for independent samples) with only minor effects on the number of assigned formulae. The noise peak removal performance was further improved by excluding outlier spectra and less frequently occurring peaks via the PSRSS workflow. Moreover, prediction of electron-denoting capacity in swimming pool water by FT-ICR MS spectra was significantly improved (p < 0.05, ANOVA) by the PSRSS workflow. Overall, our findings demonstrated the feasibility of the PSRSS workflow to serve as a robust, labor-saving method to remove noise peaks and cluster spectra, yielding reliable spectral information for multiple independent FT-ICR MS spectra.

暂无