Non-target screening to identify unknowns, Automation and increasing confidence

The reliable identification of an unknown micro-pollutant in water is not only essential to good (human) risk assessment, it is also necessary to predict the behavior of a substance in the environment and in drinking water treatment. In 2005 a breakthrough took place in the investigation of the presence of organic micro-pollutants in water with the introduction of high resolution mass spectrometry (HRMS). Combined with liquid chromatography (LC), screening
is thus carried out at low concentration levels (ng / L range) and for a wide range of substances, typically referred to as LC-HRMS based non-target screening (NTS).
Suspect screening can be applied to NTS data to screen for candidate substances which are suspected and/or expected to be present in a sample. This is done on the basis of data from databases, including exact mass, isotope pattern, MS2 fragmentation pattern and metadata (McEachran et al., 2017; Schymanski and Williams, 2017). If a substance is missing in available databases – as is often the case with transformation products – then manual identification is required. Depending on how certain the identification, the identified structure is provided with a defined level of confidence (Schymanski et al., 2014). In the BTO project “Mass Spectrometry: tools for unknown IDs”, a workflow was developed describing these steps (BTO 2017.073). From that project it became clear that NTS data interpretation is a time- and labor- intensive process that requires a lot of expertise and manual work from the researcher (see document TG NMS 15-04-06). In addition, a trajectory towards a Dutch technical agreement to eventually substantiate a legal standard for NTS screening was started, which requires streamlined non- target data interpretation and identification with high reliability.

De betrouwbare identificatie van een onbekende microverontreiniging in water is essentieel voor de risicobeoordeling en het voorspellen van het gedrag van de verontreiniging in het milieu en in de drinkwaterzuivering. Om onbekende microverontreinigingen sneller en met een hogere betrouwbaarheid te kunnen identificeren, zijn twee geautomatiseerde workflows ontwikkeld voor non-target screening data-analyse; een is gebaseerd op het open source software package patRoon, de andere op de commerciële software Compound Discoverer. De twee workflows zijn vervolgens gebruikt om data van KWR en de drinkwaterbedrijven te analyseren. In een hands-on data-analyseworkshop bij KWR hebben medewerkers van drinkwaterlaboratoria de workflows met succes toegepast.