Tracing seafood at high spatial resolution using NGS-generated data and machine learning: Comparing microbiome versus SNPs

Abstract

Developing reliable tools to trace food origin represents a major goal for producers and control authorities. Here, we test the hypothesis whether NGS-generated data could provide a reliable tool to ensure seafood traceability. As a test case, we used the Manila clam Ruditapes philippinarum, a bivalve mollusk of high commercial interest with worldwide distribution, collected in the Venice lagoon sites subjected to prohibition of clam harvesting because of chemical contamination as well as in authorized clam harvesting areas. The results obtained demonstrated that the geographic origin of Manila clam may be more accurately determined basing on microbiome data than single nucleotide polymorphisms. In particular, combining microbiome data with machine-learning techniques, we provide the experimental evidence that it is possible to trace the clam place of origin at high spatial resolution. Considering its low cost and portability, NGS-analysis of microbiome data might represent a cost-effective, high-resolution tool for reliable food traceability.

Keywords: Food traceability; Machine learning; Manila clam Ruditapes philippinarum; Microbiome; NGS; Shellfish; Venice lagoon.