Processing of non-targeted LC-HRMS data: Progress and challenges in automated data-driven optimisations

Liquid chromatography coupled to high-resolution mass spectrometry (LC-HRMS) has become a cornerstone analytical technique for non-targeted chemical profiling of complex environmental, biological, and food samples. However, processing non-targeted LC-HRMS data remains challenging due to numerous parameters that need to be optimised and inaccurate, inconsistent feature detection across software platforms, which might lead to false discoveries and missed compound detections. This critical review provides a detailed description of commonly used algorithms for feature detection and retention time alignment, including their strengths and weaknesses, comparative performance evaluations, and recent advancements in these two key processing steps. Moreover, practical guidelines for manual tuning of key parameters in selected tools are provided, along with a critical evaluation of both established and recent software designed to filter false-positive features, assess data-processing accuracy, and automatically optimise parameter settings. We emphasise the importance of adopting a robust and reproducible data processing pipeline involving a multi-tool strategy that combines: (1) performance checks to ensure instrument and data quality; (2) objective QA/QC evaluation metrics to assess tool performance; (3) optimisation tools validated for specific sample types and instruments; (4) filtering tools to reduce false positives; and (5) comprehensive reports with actionable criteria to improve data processing outcomes. Integrating this modular pipeline into an open-source, version-controlled repository with transparent reporting standards will be essential for ensuring accurate, reliable and reproducible outcomes, particularly in fields like food safety and human and environmental monitoring, where low contaminant levels add an additional layer of complexity.