A chemotyping and genotyping comprehensive approach may be useful for the analytical traceability of food ingredients. The objective of this work was developing innovative models for discerning between L. albus and L. angustifolius, the most used in human nutrition, since they are not technologically equivalent. In this context, we developed two models applying multivariate statistical analysis (Principal Component Analysis, PCA) and artificial intelligence (Self Organizing Maps, SOM’s) onto chemical parameters (proximate composition, alkaloids, tocopherols) or Random Polymorphic DNA fingerprints. Additionally, we applied, for the first time on lupin, a Kohonen’s Artificial Neural Network, including both the chemical and the genetic data, which appeared to be a powerful approach for clustering not only species, but also cultivars, and added some new information about their genetic similarity. The possibility of discriminating L. albus from L. angustifolius is relevant for lupin traceability; the foreseen fields of application are refined flours or ingredients, where morphological analysis is not applicable.
Discrimination of genotype/chemotype of Lupinus albus and Lupinus angustifolius by the artificial intelligence-based chemometrical characterization / A. Arnoldi, J.D. Coisson, M. Arlorio, M. Locatelli, C. Garino, D. Resta, E. Sirtori, G. Boschin - In: Abstract of the 13th International Lupin Conference / [a cura di] B. NAGANPWSKA, P. KACHLICKI, B. WOLKO. - Canterbury : International Lupin Association, 2011 Jun 06. - ISBN 9788361607731. - pp. 36-36 (( Intervento presentato al 13. convegno International Lupin Conference tenutosi a Poznań nel 2011.
Discrimination of genotype/chemotype of Lupinus albus and Lupinus angustifolius by the artificial intelligence-based chemometrical characterization
A. Arnoldi;D. Resta;E. Sirtori;G. Boschin
2011
Abstract
A chemotyping and genotyping comprehensive approach may be useful for the analytical traceability of food ingredients. The objective of this work was developing innovative models for discerning between L. albus and L. angustifolius, the most used in human nutrition, since they are not technologically equivalent. In this context, we developed two models applying multivariate statistical analysis (Principal Component Analysis, PCA) and artificial intelligence (Self Organizing Maps, SOM’s) onto chemical parameters (proximate composition, alkaloids, tocopherols) or Random Polymorphic DNA fingerprints. Additionally, we applied, for the first time on lupin, a Kohonen’s Artificial Neural Network, including both the chemical and the genetic data, which appeared to be a powerful approach for clustering not only species, but also cultivars, and added some new information about their genetic similarity. The possibility of discriminating L. albus from L. angustifolius is relevant for lupin traceability; the foreseen fields of application are refined flours or ingredients, where morphological analysis is not applicable.Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.