Co-frangitura oliva e nocciola: Studio del prodotto

In recent times, the choices of consumers have been more conscious and oriented to foods with health benefits. The present paper deals with the study of oil from crushing of olive and huzelnut with the aim of obtaining a “functional food”. Different samples of oil derived from the crushing of olive (O), olive with 5% of hazelnut (O5N) and olive with 10% of hazelnut (O10N), exposed to different temperatures (28 and 35°C) and times (15 and 30 minutes) of malaxation. The samples of oil were initially subjected to a qualitative assessment by the analysis of peroxide and free acidity. Following further analyses were carried out namely the determination of fatty acids and triglycerides by FAST GC-FID, the determination of tocopherols by HPLC-FLC, the analysis of sterols by GC/MS and the spectroscopic analysis with FT-MIR combined with statistical analysis with PCA and PLS. The results showed that increasing the time and temperature of malaxation there aren’t relevant significant differences (p<0,05) in the composition of fatty acids, triglycerides and tocopherols in the different oils, but there are higher extraction yields. The increase of content of hazelnut in phase of crushing causes the decrease of triglycerides C50 and C52, the increase of the class C54, total tocopherols and of total sterols as well. The samples analysed with FT-MIR spectroscopy have showed, on the contrary to conventional analytical techniques, a good discrimination between different oils despite of the similar chemical composition of olive and hazelnuts. After the PLS models were built from spectra FT-MIR in order to estimate the content of triglycerides C50, C52 and C54 and total tocopherols, with good R2 in full cross validation (R2>0,821).

Applicazione del metodo QDanet_PRO alla classificazione di dati omici

Il presente lavoro di tesi si pone nell'ambito dell'analisi dati attraverso un metodo (QDanet_PRO), elaborato dal Prof. Remondini in collaborazine coi Dott. Levi e Malagoli, basato sull'analisi discriminate a coppie e sulla Teoria dei Network, che ha come obiettivo la classificazione di dati contenuti in dataset dove il numero di campioni è molto ridotto rispetto al numero di variabili. Attraverso questo studio si vogliono identificare delle signature, ovvero un'insieme ridotto di variabili che siano in grado di classificare correttamente i campioni in base al comportamento delle variabili stesse. L'elaborazione dei diversi dataset avviene attraverso diverse fasi; si comincia con una un'analisi discriminante a coppie per identificare le performance di ogni coppia di variabili per poi passare alla ricerca delle coppie più performanti attraverso un processo che combina la Teoria dei Network con la Cross Validation. Una volta ottenuta la signature si conclude l'elaborazione con una validazione per avere un'analisi quantitativa del successo o meno del metodo.