Classifying the pesticides in foods between GC-amenable and LC-amenable using the prediction model with molecular descriptors

Význam tématu

Analytical chemists frequently need to determine whether a pesticide is amenable to gas chromatography (GC/MS) or liquid chromatography (LC/MS). Accurate classification helps optimize analytical methods, improves detection reliability and streamlines laboratory workflows.

Cíle a přehled studie / článku

This study aimed to develop a quantitative structure–property relationship (QSPR) model using molecular descriptors and machine learning to predict the amenability of pesticides to GC/MS or LC/MS. It combined validated residue analysis data from FDA (136 pesticides in avocado) and EURL (127 pesticides in olive oil), resulting in 194 compounds after removing inconsistencies.

Použitá metodika a instrumentace

• Data collection: Pesticide lists from FDA and EURL validation reports.
• Chemical representation: Canonical SMILES obtained from PubChem.
• Molecular descriptors: 224 descriptors calculated via R’s rcdk package, reduced to 176 after filtering zero-variance features.
• Machine learning: 119 classification algorithms from the caret package evaluated with 10-fold cross validation.
• Performance metrics: Accuracy of CV10 resamples and execution time measured in R.

Použitá instrumentace

• Gas chromatography–mass spectrometry (GC/MS)
• Liquid chromatography–mass spectrometry (LC/MS)

Hlavní výsledky a diskuse

• The average classification accuracy across all methods was 77 %.
• Ensemble decision tree algorithms delivered the best performance.
• AdaBoost.M1 achieved the highest accuracy (85.5 %) but with long execution time (~1 h 34 min).
• xgbDART reached 85.0 % accuracy with moderate runtime (~8 min 33 s).
• xgbTree balanced speed and accuracy (84.6 % in under 2 min).
• Bagging approaches like bagEarth underperformed for this dataset.

Přínosy a praktické využití metody

This QSPR-based classification tool can guide analysts in selecting the appropriate chromatography technique, reducing trial-and-error in method development and saving time and resources in routine pesticide monitoring.

Budoucí trendy a možnosti využití

• Expansion of training sets with additional pesticides and matrices.
• Integration of more advanced descriptors and deep learning models.
• Real-time decision support systems embedded in analytical software.
• Adaptation to emerging pesticide chemistries and novel residues.

Závěr

The QSPR and machine learning framework effectively predicts pesticide amenability to GC/MS or LC/MS. Based on accuracy and runtime balance, xgbDART is recommended for practical deployment.

Reference

• Barganska Z., Konieczka P., Namieśnik J., 2018. Comparison of Two Methods for the Determination of Selected Pesticides in Honey and Honeybee Samples. Molecules 23:2582.
• Anagnostopoulos C., Miliadis G.E., 2013. Development and validation of an easy multiresidue method for the determination of multiclass pesticide residues using GC–MS/MS and LC–MS/MS in olive oil and olives. Talanta 112:1–10.
• Food and Drug Administration, Pesticide Analytical Manual Vol. I, Appendix II, 1999.
• Chamkasem N., Ollis L.W., Harmon T., Lee S., Mercer G., 2013. Analysis of 136 Pesticides in Avocado Using a Modified QuEChERS Method with LC/MS/MS and GC/MS/MS. J. Agric. Food Chem. 61:2315–2329.
• EU Reference Laboratories for Residues of Pesticides, EURL-FV(2012-M6) Validation Data of 127 Pesticides Using a Multiresidue Method by LC/MS/MS and GC/MS/MS in Olive Oil, 2012.
• Kuhn M., 2008. Building Predictive Models in R Using the caret Package. J. Stat. Softw. 28:1–26.