| Abstract Title: | Predicting Ionization Efficiency in Non-Targeted Analysis: A Novel Approach Using Molecular Fingerprints and Cumulative Neutral Loss Probability |
| Presenter Name: | Mr Alexandros Nikolopoulos |
| Co-authors: | Mrs Denice van Herwerden Mrs Viktoriia Turkina Dr Saer Samanipour |
| Company/Organisation: | University of Amsterdam |
| Country: | Netherlands |
Abstract Information :
One of the most important challenges of non-targeted analysis is the quantification of the analytes of interest. Due to the lack of internal standards and calibration curves, semi-quantitative approaches are used. The challenge arises from the electrospray ionization process prior to mass spectrometry analysis. Specifically, the ionization efficiency (IE) of different compounds varies significantly and therefore the relative peak areas may differ from the relative concentrations. The aim of such approaches is to predict the ionization efficiency of the analyte of interest and find its correlation with the response factor. With this information, the analyte concentration can be calculated using the peak area and the response factor. Previous studies have investigated the use of descriptors for the prediction of IE. In this study, we present a novel approach for predicting ionization efficiency using various types of molecular fingerprints and comparing their potential. Another important point of study is the investigation of the ability to predict IE based on cumulative neutral loss (CNL) probabilities, since structural information of the molecule of interest is not required. Our method employs random forest regressors to analyze the fingerprints and CNLs, resulting in an accurate prediction of ionization efficiency. By providing a more accurate prediction of ionization efficiency, this approach can help enhance the overall performance of non-targeted analysis and provide a more reliable method of quantifying analytes.
