P27 | NGS technologies supporting official food control activities: comparison of metabarcoding analysis on Illumina and Ion Torrent platforms and the official method (DNA microarray) applied to the authentication of ethnic products

A. Giusti¹, N. Lombardo¹, P. De Santis², B.M. Varcasia², U. Marchesi², P. Marconi², L. Gasperetti², A. Armani¹ | ¹Dipartimento di Scienze Veterinarie, Università di Pisa; ²Istituto Zooprofilattico Sperimentale del Lazio e della Toscana M. Aleandri, Roma, Italy

Abstract

Purpose. Within the framework of official control activities aimed at food authentication in Italy, samples taken nationally by the Competent Authority (CA) are sent to the territorially competent Experimental Zooprophylactic Institute (IZS). The official analysis for species identification is based on a validated low-density DNA microarray (MBD) technique. This technique, designed to detect a limited number of target species, may have limitations in the analysis of complex food products containing non-conventional or unexpected species, as is frequently found in ethnic foods. For this reason, the approach to detecting fraud in multi-species products is increasingly shifting toward methods based on Next Generation Sequencing (NGS) technologies. In this study, two NGS platforms – Illumina (ILM) and Ion Torrent (ION) – were used to analyze, using a metabarcoding approach, 56 samples from ethnic foods (already authenticated by the IZS of Lazio and Tuscany via MBD). The results obtained were compared with those of the reference method in order to evaluate the applicability of NGS technologies as a screening method for the official analysis of complex food products. Methods. A comparative analysis using the MBD method was conducted for both platforms regarding the results of all samples. The analysis included: 1) the distribution of concordance, where each sample was classified based on the degree of qualitative agreement between the species detected by the tested methods and the MBD. The categories considered are: complete concordance (identical lists), partial concordance (at least one species in common but with differences), and no concordance (no species in common). Frequency and percentages were calculated for each category; 2) the performance metrics related to: Relative sensitivity (rs) was calculated as the proportion of species detected by MBD that were also identified by ILM or ION, while the overcalling rate (or) estimated the proportion of species detected exclusively by ILM or ION and not confirmed by the reference method. Results. 1) Concordance Distribution: ILM showed complete agreement in 16 samples (28.5%), partial agreement in 31 samples (55.4%), and no agreement in 9 samples (16.1%). ION showed complete agreement in 29 samples (51.8%), partial agreement in 15 samples (26.8%), and disagreement in 12 samples (21.4%). 2) Relative performance metrics: The response rate was 68.2% in ILM and 66.4% in ION, while the OR was 27.5% in ILM and 10.8% in ION. Conclusions. The results indicate that ILM has a higher species detection capacity, but also a higher risk of overestimation or unconfirmed identifications. Conversely, ION shows greater sample-level agreement and a lower limit of detection, at the expense of slightly reduced sensitivity. These results highlight the importance of considering sensitivity and specificity together when interpreting metabarcoding data, especially when applied in a regulatory context for official food control. It should be considered that both the official method (MBD) and ILM and ION provide only qualitative data, with possible biases in species detection. For this reason, the integration of quantitative approaches could be a useful support in the context of official control.