This study investigates the metabolome of high-quality hazelnuts (Corylus avellana L.) by applying untargeted and targeted metabolome profiling techniques to predict industrial quality. Utilizing comprehensive two-dimensional gas chromatography and liquid chromatography coupled with high-resolution mass spectrometry, the research characterizes the non-volatile (primary and specialized metabolites) and volatile metabolomes. Data fusion techniques, including low-level (LLDF) and mid-level (MLDF), are applied to enhance classification performance. Principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA) reveal that geographical origin and postharvest practices significantly impact the specialized metabolome, while storage conditions and duration influence the volatilome. The study demonstrates that MLDF approaches, particularly supervised MLDF, outperform single-fraction analyses in predictive accuracy. Key findings include the identification of metabolites patterns causally correlated to hazelnut’s quality attributes, of them aldehydes, alcohols, terpenes, and phenolic compounds as most informative. The integration of multiple analytical platforms and data fusion methods shows promise in refining quality assessments and optimizing storage and processing conditions for the food industry.
Artificial intelligence decision making tools in food metabolomics: Data fusion unravels synergies within the hazelnut (Corylus avellana L.) metabolome and improves quality prediction
Squara, SimoneFirst
;Caratti, Andrea;Fina, Angelica;Liberto, Erica;Castello, Giuseppe;Bicchi, Carlo;Cordero, Chiara
2024-01-01
Abstract
This study investigates the metabolome of high-quality hazelnuts (Corylus avellana L.) by applying untargeted and targeted metabolome profiling techniques to predict industrial quality. Utilizing comprehensive two-dimensional gas chromatography and liquid chromatography coupled with high-resolution mass spectrometry, the research characterizes the non-volatile (primary and specialized metabolites) and volatile metabolomes. Data fusion techniques, including low-level (LLDF) and mid-level (MLDF), are applied to enhance classification performance. Principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA) reveal that geographical origin and postharvest practices significantly impact the specialized metabolome, while storage conditions and duration influence the volatilome. The study demonstrates that MLDF approaches, particularly supervised MLDF, outperform single-fraction analyses in predictive accuracy. Key findings include the identification of metabolites patterns causally correlated to hazelnut’s quality attributes, of them aldehydes, alcohols, terpenes, and phenolic compounds as most informative. The integration of multiple analytical platforms and data fusion methods shows promise in refining quality assessments and optimizing storage and processing conditions for the food industry.File | Dimensione | Formato | |
---|---|---|---|
final full text.pdf
Accesso aperto
Descrizione: full text
Tipo di file:
PDF EDITORIALE
Dimensione
4.81 MB
Formato
Adobe PDF
|
4.81 MB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.