Metabolic fingerprinting-based multiclass strategy for varietal authentication of wine using advanced data mining and chemometric tools

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F60461373%3A22330%2F21%3A43922876" target="_blank" >RIV/60461373:22330/21:43922876 - isvavai.cz</a>

Nalezeny alternativní kódy

RIV/60461373:22330/21:43922879

Výsledek na webu

—

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Metabolic fingerprinting-based multiclass strategy for varietal authentication of wine using advanced data mining and chemometric tools

Popis výsledku v původním jazyce

One of the major challenges in wine analysis is authentication of grape variety from which the wine is produced. Here, we have developed and validated an authentication method based on metabolomic fingerprinting of pure wine (without prior extraction), using ultra-high-performance liquid chromatography coupled with quadrupole orbitrap high resolution mass spectrometry. In total, 101 red wine and 97 white wine samples (five varieties each group) were analysed within our study. After the data mining and data pre-treatment steps, principal component analysis (PCA) was used to explore the data structure. Since the obtained unsupervised PCA models revealed a notable clustering according to the grape varieties, orthogonal partial least squares discriminant analysis (OPLS-DA) was used to create supervised binary models. To obtain the best models possible, an in-house R script for optimization of data filtration parameters was developed and implemented. From over 85 000 models calculated, final models with the highest prediction abilities were chosen, validated and arranged into a Rooted Binary Decision Directed Acyclic Graph (RBDDAG), which was used for wine authentication. In case of white wine varieties, over 96 % of samples in negative ionization mode (ESI-) and over 95 % of samples in positive ionization mode (ESI+) were correctly classified. In case of red wines, over 95 % of samples in ESI+ and over 94 % of samples in ESI- were correctly classified. Worth mentioning that the aforementioned accuracies are calculated for the whole RBDDAG structure. Multiple individual binary models achieved prediction ability as high as 100%, however, for some of the binary models present in the RBDDAG structure, correct classification of the grape variety appeared to be more problematic due to a higher genetic (and therefore metabolomic) similarity between varieties. Our results indicate that our method is suitable for official control purposes.

Název v anglickém jazyce

Metabolic fingerprinting-based multiclass strategy for varietal authentication of wine using advanced data mining and chemometric tools

Popis výsledku anglicky

One of the major challenges in wine analysis is authentication of grape variety from which the wine is produced. Here, we have developed and validated an authentication method based on metabolomic fingerprinting of pure wine (without prior extraction), using ultra-high-performance liquid chromatography coupled with quadrupole orbitrap high resolution mass spectrometry. In total, 101 red wine and 97 white wine samples (five varieties each group) were analysed within our study. After the data mining and data pre-treatment steps, principal component analysis (PCA) was used to explore the data structure. Since the obtained unsupervised PCA models revealed a notable clustering according to the grape varieties, orthogonal partial least squares discriminant analysis (OPLS-DA) was used to create supervised binary models. To obtain the best models possible, an in-house R script for optimization of data filtration parameters was developed and implemented. From over 85 000 models calculated, final models with the highest prediction abilities were chosen, validated and arranged into a Rooted Binary Decision Directed Acyclic Graph (RBDDAG), which was used for wine authentication. In case of white wine varieties, over 96 % of samples in negative ionization mode (ESI-) and over 95 % of samples in positive ionization mode (ESI+) were correctly classified. In case of red wines, over 95 % of samples in ESI+ and over 94 % of samples in ESI- were correctly classified. Worth mentioning that the aforementioned accuracies are calculated for the whole RBDDAG structure. Multiple individual binary models achieved prediction ability as high as 100%, however, for some of the binary models present in the RBDDAG structure, correct classification of the grape variety appeared to be more problematic due to a higher genetic (and therefore metabolomic) similarity between varieties. Our results indicate that our method is suitable for official control purposes.

Druh

O - Ostatní výsledky

CEP obor

—

OECD FORD obor

10406 - Analytical chemistry

Projekt

<a href="/cs/project/LM2018100" target="_blank" >LM2018100: Infrastruktura pro propagaci metrologie v potravinářství a výživě v České republice</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)<br>S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2021

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů