Estimation of the percentage of transgenic Bt maize in maize flour mixtures using perfusion and monolithic reversed-phase high-performance liquid chromatography and chemometric tools

The estimation of the percentage of transgenic Bt maize in maize flour mixtures has been achieved in this work by high-performance liquid chromatography using perfusion and monolithic columns and chemometric analysis. Principal component analysis allowed a preliminary study of the data structure. Then, linear discriminant analysis was used to develop decision rules to classify samples in the established categories (percentages of transgenic Bt maize). Finally, linear regression (LR) and multivariate regression models (namely, principal component analysis regression (PCR), partial least squares regression (PLS-1), and multiple linear regression (MLR)) were assayed for the prediction of the percentages of transgenic Bt maize present in a maize flour mixture. Using the relative areas of the protein peaks, MLR provided the best models and was able to predict the percentage of transgenic Bt maize in flour mixtures with an error of ±5.3%, ±2.3%, and ±3.8% in the predictions of Aristis Bt, DKC6575, and PR33P67, respectively.     

Rapid characterisation of (glyphosate tolerant) transgenic and non-transgenic soybeans using chromatographic protein profiles

The characterisation of transgenic and non-transgenic soybeans was performed in this work based on the examination of their protein profiles obtained by rapid chromatographic techniques. Two reversed-phase chromatographic methods using monolithic and perfusion stationary phases were applied to the separation of soybean proteins from different transgenic and non-transgenic soybeans. The development of the monolithic LC methodology was carried out through the study of the influence of different parameters such as gradient, ion-pairing reagent, and temperature on the separation of soybean proteins. Results from monolithic LC analysis were compared with those obtained by perfusion LC using a method previously developed by our research team. Perfusion and monolithic LC enabled the separation of soybean proteins in less than 3 and 8 min, respectively. In both cases, there were certain features in the chromatograms that seemed to be characteristic of transgenic samples. A deeper analysis of chromatographic profiles was then performed by the application of multivariate classification techniques. Results from these multivariate techniques showed that the two methods presented similar classification capabilities being both suitable for the characterisation of transgenic and non-transgenic soybeans. A more robust mathematical model was built with the data obtained by the perfusion method using 10 additional samples for training (a total of 26 samples), obtaining a 96.2% of correct classification. This model was validated by a cross-validation procedure (80.8% of correct classification) and by the correct classification of 15 out of 16 blind samples.     

Exploratory chemometric analysis for the characterisation of Tunisian olive cultivars according to their lipid and sterolic profiles

This study aims to evaluate some Tunisian olive genetic resources using fatty acid, triacyglycerol and sterolic compositions and to classify the cultivars according to their fruit genotype and to their respective geographical origin (North, Centre and South). manova results showed that the studied cultivars presented highly significant differences regarding all the variables (P < 0.01). The most discriminant variables of fatty acids are C17:1 (F = 98.468), C16:0 (F = 92.994), C18:1 (F = 60.865), C18:1/C18:2 (F = 44.632) and C18:2 (F = 40.167); those of triacylglycerols are POP (F = 123.34), LLL (F = 122.944), LnLO (F = 98.363), POO (F = 93.357) and LOO (F = 90.42), while sitostanol (F = 289.171), campestanol (F = 192.792) and campesterol (F = 160.724) have the higher discriminant power among sterol compounds. Principal component analysis (PCA) was performed on the data of each chemical parameter to explore their usefulness for the discrimination of eleven monovarietal olive oils. Best differentiations among cultivars were obtained with triacyglycerol and sterolic compositions. The spatial distribution of the different oil samples using all the collected data showed a good discrimination among olive cultivars. A strong resolution between the samples according to the geographical origin was obtained by means of factorial discriminant analysis (FDA) (λ = 0.002). Comparisons of the distances between classes were statistically significant (Fisher tests; P < 0.0001), and 90.91% of cross‐validated grouped cases are correctly classified. The obtained results could become an important tool for sorting out oils to a single cultivar or to a specific geographical area.