| Abstract Title: | Optimization of Compound Discoverer® Quality Parameters to Ease Peak-Picking Steps in Non-target Screening of Xenobiotics in Human Biofluids |
| Presenter Name: | Ms Inés Baciero |
| Co-authors: | Mr Mikel Musatadi Prof Maitane Olivares Prof Ailette Prieto Prof Olatz Zuloaga |
| Company/Organisation: | University of the Basque Country (UPV/EHU) |
| Country: | Spain |
Abstract Information :
Non-target screening has gained relevance due to the need of broad-scoped analysis regarding environmental samples, exposome studies or metabolomics, to name a few. Compound Discoverer® (CD) is one of the available commercial softwares to extract information for such complex tasks. Nonetheless, the peak-picking step is still a main bottleneck when interpreting results. There, CD automatically selects the “peaks” (features) which may correspond to a compound of interest, however it tends to consider as optimal signals that are fundamentally noise or baseline, even if other filter-thresholds are applied. Consequently, it is currently required a highly time-consuming manual check of all the features selected, enhancing the possibilities of making human errors during the systematic revision of thousands of items. Recently, a new update for CD implemented Peak Quality Factor (PQF) filters, which consider peak-shape characteristics when retrieving features. Taking the above mentioned into consideration, the aim of the present work was to reach consensus PQF values included in the latest version of CD (3.3), to ease non-target screening regarding xenobiotic analysis in human biofluids. For that task, milk and urine blanks and xenobiotic-spiked (up to 187 compounds and metabolites) human urine and breast milk samples were analysed with previously validated analytical procedures based on liquid-chromatography coupled to high-resolution mass-spectrometry (LC-HRMS). The data obtained was later processed in CD and PQF parameters were optimized. At first, all filters except the signal to noise ratio were disabled, and the features corresponding to the spiked xenobiotics were identified and tagged. Then, while maintaining the tag as true, available PQFs (FWHM2Base, Zig-zag index, Jaggedness and Modality) were enabled. Individual optimal values were adjusted to find the threshold where the maximum amount of the xenobiotics was identified, while minimizing the retrieved feature amount. Consensus values for both positive and negative ionization modes were established where the detected analyte proportion vs. PQF values’ curve reached a plateau. These findings entail a preliminary approach for the equilibrium between obtaining trustworthy results and required time consumption. Additionally, it will allow the reduction of erroneous conclusions in the non-target analysis of xenobiotics in human biofluids.
