Integration of non-target screening, statistical analyses and bioassays for the global assessment of chemical water quality

Thanks to developments in screening methods based on high-resolution mass spectrometry (HRMS), the detection of chemicals in water now not only allows for the detection of specific substances (target substance analysis) but also for broader screening (non-target screening). Non-target screening is a promising tool for the evaluation of changes in chemical water quality during water treatment. However, because this method generates very large amounts of data, the structural identification of all the ‘peaks’ found is virtually impossible.

Furthermore, transformation products are often missing from the available databases but they can be predicted using known transformation rules. As an alternative, data science methods can be used to identify differences between water treatment steps. It is therefore important to investigate how, and how meaningfully, different levels of data generated with non-target screening can be used to evaluate, for example, the effectiveness of water treatment. In non-target screening, we distinguish between three different levels of generated information: from “unknown peaks” to “suspects” (peaks on a known “suspect list”) and “trend profiles” (data patterns that result from the combination of the previous levels). In addition, information emerges from bioassays on the basis of effects. Suspect screening was improved and accelerated by retrieving chemical characteristics and detailed suspect lists online. By integrating data from the chemical non-target screening with the biological effect data, toxicity was also included in the prioritisation and it was possible to establish a more comprehensive picture of chemical water quality.

The study has resulted in a range of tools and work flows that have been successfully applied in the case studies to provide an even more comprehensive picture of chemical water quality.

  • The semi-automatic generation of suspect lists for transformation products
  • Statistical tool set for the interpretation of data from non-target screening
  • A set of non-target trend profiles associated with drinking water treatment steps
  • New work flows for non-target screening and bioassay data interpretation in R scripts
  • Lists of detected suspects, including transformation products, for subsequent structural elucidation