How can DNA-based omics effectively improve microbiological quality control? An explorative study applied to infant food production

Community profiling genomics, metagenomics and pangenomics are nowadays extensively used in medical research to obtain a holistic view of the patient's condition and to highlight biomarkers associated with human pathologies and diseases. Although they represent well-established techniques, in food science and technology their use is still marginal and their potentialities are far from being fully understood and thus effectively exploited by industries in microbiological control. It is therefore pivotal to explore their applicability on site, starting with food processes in which safety risks lead to the highest impact on certain population, such as the production of infant food. Therefore, a comprehensive monitoring of microbiota and microbiome has been conducted during the production of a cereal based powder for fourteen months, sampling the environment, raw materials, intermediate and final products. Total DNA of each sample was subjected to 16S rRNA amplicon-based and shotgun sequencing. In addition to these untargeted approaches, targeted detection by enrichment was performed for five microorganisms: Bacillus cereus, Clostridium perfringens, Staphylococcus aureus, Salmonella spp. and Listeria monocytogenes. The presence of alive cells was confirmed by qPCR and isolation, which was followed by pangenome analyses. Resident environmental microbiota and microbiome changed over time, being affected by different incoming raw materials of each production run, in a frame of powder foodstuffs and high airborne contaminations. Both targeted and untargeted approaches showed a high prevalence of Bacillus cereus, mainly in raw materials. Furthermore, the modelling of the production processes through network analysis allowed the determination of potential contamination pathways for all five microorganisms researched, which appeared to be invisible to direct linear observations. Interestingly, the hindering effect of not-bacterial DNA was less evident in community profiling data than metagenomic data, which however provided a better taxonomic resolution of minor taxa in samples with high DNA quality. Finally, the isolate-dependent nature of pangenomic data was pivotal for the discrimination between resident and transient biotypes within the processing plant. Regardless of the process-specific outcomes, this in situ investigation underlines the imperative need to find thoughtful methods that simplify the complexity of omics output to allow better interpretation and application in food industries.