In-Depth Analysis of the Data from an Interlaboratory Study of Quantitative Non-Target Screening-How Do the...

Area of Science:

• Environmental Science
• Analytical Chemistry
• Computational Chemistry

Background:

• Non-target screening with liquid chromatography-high-resolution mass spectrometry is crucial for environmental monitoring.
• Quantifying detected suspected contaminants remains a significant challenge, prompting the development of various approaches.

Purpose of the Study:

• To analyze interlaboratory comparison data on quantification approaches for environmental contaminants.
• To investigate the link between prediction errors and instrument parameters in machine learning-based quantification.
• To assess the comparability of response factors (RFs) across different datasets and instrumental method limitations.

Main Methods:

• Analysis of data from a previous interlaboratory comparison study.
• Evaluation of machine learning-based quantification leveraging predicted ionization efficiencies.
• Investigation of response factor (RF) comparability using linear models for scaling across datasets.

Main Results:

• No specific instrument parameters were definitively linked to systematic prediction errors.
• The choice of organic modifier and/or additive type impacted the detection of certain compounds.
• Comparable logRFs were achieved across datasets after linear projection, but with compression for dissimilar datasets.
• Compounds with lower logRF exhibited greater variability across datasets.

Conclusions:

• Machine learning approaches using predicted ionization efficiencies show strong potential for contaminant quantification.
• Instrumental method choices, particularly organic modifiers and additives, can influence detection.
• Data scaling and compound-specific variability are key considerations for robust quantification across diverse datasets.