Influence of extraction windows for data-independent acquisition on feature annotation during suspect screening

Abstract

Non-target analysis (NTA) using high-resolution mass spectrometry is becoming a useful approach to screen for suspect and unknown chemicals. For comprehensive analyses, data-independent acquisition (DIA), like Sequential Windowed Acquisition of all THeoretical Mass Spectra (SWATH-MS) on Sciex instruments, is necessary, usually followed by library matching for feature annotation. The choice of parameters, such as acquisition window number and size, may influence the comprehensiveness of the suspect features detected. The goal of this study was to assess how mass spectrometric DIA settings may influence the ability to obtain confident annotations and identifications of features in environmental (river water, passive sample extract (PSE)), wastewater (unpreserved and acidified) and biological (urine) sample matrices. Each matrix was analysed using 11 different MS methods, with 5–15 variable size acquisition windows. True positive (TP) annotation (i.e., matching experimental and library spectra) rates were constant for PSE (40%) and highest for urine (18%), wastewater (34% and 36%, unpreserved and acidified, respectively) and river water (8%) when using higher numbers of windows (15). The number of annotated features was highest for PSE (12%) and urine (8.5%) when using more acquisition windows (9 and 14, respectively). Less complex matrices (based on average total ion chromatogram intensities) like river water, unpreserved and acidified wastewater have higher annotation rates (7.5%, 8% and 13.2%, respectively) when using less acquisition windows (5–6), indicating matrix dependency of optimum settings. Library scores varied widely for correct (scores between 6 and 100) as well as incorrect annotations (scores between 2 and 100), making it hard to define specific ideal cut-off values. Results highlight the need for properly curated libraries and careful optimization of SWATH-MS and other DIA methods for each individual matrix, finding the best ratio of total annotations to true positive, (i.e., correct) annotations to achieve best NTA results.