Harmonized quality assurance/quality control provisions to assess completeness and robustness of MS1 data preprocessing for LC-HRMS-based suspect screening and non-targeted analysis#

Authors#

Sarah Lennon, Jade Chaker, Marja Lamoree, Carolin Huber, Julien Parinet, Thierry Guérin, Tobias Schulze, Nicolas Creusot, Michael A. Stravs, Rosalie Nijssen, Žiga Tkalec, Baninia Habchi, Jean-Philippe Antignac, Emilien L. Jamin, Nafsika Papaioannou, Katerina Gabriel, Dimosthenis Sarigiannis, Emma Schymanski, Lutz Ahrens, Tina Kosjek, Elliott J. Price, Arthur David

Abstract#

Non-targeted and suspect screening analysis using liquid chromatography coupled to high-resolution mass spectrometry holds great promise to more comprehensively characterize real-life complex chemical exposures mixtures. However, large-scale implementation of these strategies for research and regulatory purposes still faces several analytical and informatic challenges. Data preprocessing (feature extraction from raw data) is a crucial part of the process, with many approaches described for metabolomics and exposure-related applications. However, some limitations are observed: (i) peak-picking and feature extraction might be incomplete, especially for low abundant compounds, and (ii) limited reproducibility has been observed between laboratories and software for detected features (potentially due to mass and retention time alignment or detection issues) and their relative quantification. To address the current limitations related to data preprocessing steps, we first conducted a critical literature review of existing solutions that could improve the reproducibility of data preprocessing and the efficiency of extraction of useful information from LC-HRMS data. Solutions include providing repositories and reporting guidelines, open and modular processing workflows, public benchmark datasets, tools to optimize the data preprocessing and to filter out false positive detections. Since we identified a lack of consistent and harmonized quality assurance/quality control (QA/QC) guidelines available to assess the quality of data preprocessing steps, we propose and discuss harmonized QA/QCs sets for data preprocessing relevant for human biomonitoring (HBM), food and environment communities, to ensure robust and reproducible detection of contaminants of emerging concern. The sets of harmonized QA/QC provisions include criteria that would allow to assess the sensitivity of feature detection, reproducibility, integration accuracy, precision, accuracy, and consistency of data preprocessing. Further collaborative studies will be needed to determine thresholds and tolerances for these QA/QCs.