HDPairFinder: A data processing platform for hydrogen/deuterium isotopic labeling-based nontargeted analysis of trace-level amino-containing chemicals in environmental water

• Published in: Journal of environmental sciences (China) 2024-02, Vol.136, p.583-593
• Main Authors: Zhao, Tingting, Carroll, Kristin, Craven, Caley B., Wawryk, Nicholas J.P., Xing, Shipei, Guo, Jian, Li, Xing-Fang, Huan, Tao
• Format: Article
• Language: English
• Subjects: algorithms; bioinformatics; deuterium; HPLC-HRMS; Isotope labeling; liquid chromatography; mass spectrometry; Nontargeted analysis; Reactive N-compounds; solid phase extraction; Source water; water analysis
• ISSN: 1001-0742; 1878-7320
• DOI: 10.1016/j.jes.2023.02.033

•A data processing platform, HDPairFinder, was developed to automatically recognize H/D isotope labeled chemicals in HPLC-HRMS data.•Cross-correlation was proposed to address the RT shift caused by the deuterium isotopic effect.•A suite of bioinformatic algorithms was incorporated to effectively remove false positive features.•AMINES, a library of over 38000 amino-containing chemicals, was constructed to facilitate compound annotation.•Over 1000 highly confident H/D-labeled amino-containing compounds can be identified from the data of environmental water samples. The combination of hydrogen/deuterium (H/D) formaldehyde-based isotopic methyl labeling with solid-phase extraction and high-performance liquid chromatography–high resolution mass spectrometry (HPLC-HRMS) is a powerful analytical solution for nontargeted analysis of trace-level amino-containing chemicals in water samples. Given the huge amount of chemical information generated in HPLC-HRMS analysis, identifying all possible H/D-labeled amino chemicals presents a significant challenge in data processing. To address this, we designed a streamlined data processing pipeline that can automatically extract H/D-labeled amino chemicals from the raw HPLC-HRMS data with high accuracy and efficiency. First, we developed a cross-correlation algorithm to correct the retention time shift resulting from deuterium isotopic effects, which enables reliable pairing of H- and D-labeled peaks. Second, we implemented several bioinformatic solutions to remove false chemical features generated by in-source fragmentation, salt adduction, and natural 13C isotopes. Third, we used a data mining strategy to construct the AMINES library that consists of over 38,000 structure-disjointed primary and secondary amines to facilitate putative compound annotation. Finally, we integrated these modules into a freely available R program, HDPairFinder.R. The rationale of each module was justified and its performance tested using experimental H/D-labeled chemical standards and authentic water samples. We further demonstrated the application of HDPairFinder to effectively extract N-containing contaminants, thus enabling the monitoring of changes of primary and secondary N-compounds in authentic water samples. HDPairFinder is a reliable bioinformatic tool for rapid processing of H/D isotopic methyl labeling-based nontargeted analysis of water samples, and will facilitate a better understanding of N-containing chemical compounds in water. [D