Spotlight on mass spectrometric non‐target screening analysis: Advanced data processing methods recently communicated for extracting, prioritizing and quantifying features

Abstract

Non‐target screening of trace organic compounds complements routine monitoring of water bodies. So‐called features need to be extracted from the raw data that preferably represent a chemical compound. Relevant features need to be prioritized and further be interpreted, for instance by identifying them. Finally, quantitative data is required to assess the risks of a detected compound. This review presents recent and noteworthy contributions to the processing of non‐target screening (NTS) data, prioritization of features as well as (semi‐) quantitative methods that do not require analytical standards. The focus lies on environmental water samples measured by liquid chromatography, electrospray ionization and high‐resolution mass spectrometry. Examples for fully‐integrated data processing workflows are given with options for parameter optimization and choosing between different feature extraction algorithms to increase feature coverage. The regions of interest‐multivariate curve resolution method is reviewed which combines a data compression alternative with chemometric feature extraction. Furthermore, prioritization strategies based on a confined chemical space for annotation, guidance by targeted analysis and signal intensity are presented. Exploiting the retention time (RT) as diagnostic evidence for NTS investigations is highlighted by discussing RT indexing and prediction using quantitative structure‐retention relationship models. Finally, a seminal technology for quantitative NTS is discussed without the need for analytical standards based on predicting ionization efficiencies.

Critical Assessment of the Chemical Space Covered by LC–HRMS Non-Targeted Analysis

“…One of the last steps of CEC identification is to use a database to relate the MS output to a known chemical structure. To proceed with the identification, the experimental data first undergo the following preprocessing steps: data compression, to remove noise and blank peaks; feature detection, to find features in three-dimensional data; componentization, to group together fragments and isotopologues belonging to the same compound; and feature prioritization, to reduce the number of irrelevant features . Because most of the collected studies used vendor software for the latter four steps, which makes it almost impossible to retrieve the information on the algorithms that were utilized, these parameters cannot be adequately discussed regarding their influence on the coverage of chemical space.…”

“…To proceed with the identification, the experimental data first undergo the following preprocessing steps: data compression, to remove noise and blank peaks; feature detection, to find features in three-dimensional data; componentization, to group together fragments and isotopologues belonging to the same compound; and feature prioritization, to reduce the number of irrelevant features. 81 Because most of the collected studies used vendor software for the latter four steps, which makes it almost impossible to retrieve the information on the algorithms that were utilized, these parameters cannot be adequately discussed regarding their influence on the coverage of chemical space. For the identification of known unknowns, preprocessed data are compared with chemical databases and matched against references from available spectral libraries by utilizing a combination of features, retention time, accurate mass, and fragmentation patterns.…”

Quantification Strategies for Two-Dimensional Liquid Chromatography Datasets Using Regions of Interest and Multivariate Curve Resolution Approaches

“…This broad applicability is because the detection limit is conditioned by the sensitivity of the detector, rather than by the chemometric approach itself. Hence, the versatility of the MCR-ALS method for quantification is also very appealing in a wide variety of applications where the qualitative implementation of this approach has already been demonstrated, such as in environmental analyses (e.g., pharmaceutical compounds [31][32][33], in the determination of organic matter [34], organic compounds [35] or proteins [36], both in water bodies and wastewater treatment plants, and also metabolomic studies [37][38][39]. The ROIMCR procedure has also been validated for qualitative analysis and relative quantification of LC×LC-MS datasets, in metabolomics and pharmaceutical analyses [26,27].…”

Cumulative Neutral Loss Model for Fragment Deconvolution in Electrospray Ionization High-Resolution Mass Spectrometry Data

“…Nontargeted analysis (NTA) is a growing approach to uncover the known and unknown unknowns in complex samples, containing thousands of chemical constituents. − Due to the complexity of the samples, coming from, for example, environmental or biological background, adequate data processing approaches are required for resolving the information belonging to unique chemical constituents. ,,− One of the most commonly used approaches to perform NTA is liquid chromatography coupled to high-resolution mass spectrometry (LC-HRMS). LC-HRMS, even though powerful in separating chemical constituents, is not able to fully resolve complex samples. , This lack of full separation may result in overlapping signals from multiple chemicals (e.g., matrix signal), thus overlapping features in the MS1 signal.…”

“… 1 − 8 Due to the complexity of the samples, coming from, for example, environmental or biological background, adequate data processing approaches are required for resolving the information belonging to unique chemical constituents. 1 , 4 , 7 − 21 One of the most commonly used approaches to perform NTA is liquid chromatography coupled to high-resolution mass spectrometry (LC-HRMS). LC-HRMS, even though powerful in separating chemical constituents, is not able to fully resolve complex samples.…”