Segal’s Law, 16S rRNA gene sequencing, and the perils of foodborne pathogen detection within the American Gut Project

Obtaining human population level estimates of the prevalence of foodborne pathogens is critical for understanding outbreaks and ameliorating such threats to public health. Estimates are difficult to obtain due to logistic and financial constraints, but citizen science initiatives like that of the American Gut Project (AGP) represent a potential source of information concerning enteric pathogens. With an emphasis on genera Listeria and Salmonella, we sought to document the prevalence of those two taxa within the AGP samples. The results provided by AGP suggest a surprising 14% and 2% of samples contained Salmonella and Listeria, respectively. However, a reanalysis of those AGP sequences described here indicated that results depend greatly on the algorithm for assigning taxonomy and differences persisted across both a range of parameter settings and different reference databases (i.e., Greengenes and HITdb). These results are perhaps to be expected given that AGP sequenced the V4 region of 16S rRNA gene, which may not provide good resolution at the lower taxonomic levels (e.g., species), but it was surprising how often methods differ in classifying reads—even at higher taxonomic ranks (e.g., family). This highlights the misleading conclusions that can be reached when relying on a single method that is not a gold standard; this is the essence of Segal’s Law: an individual with one watch knows what time it is but an individual with two is never sure. Our results point to the need for an appropriate molecular marker for the taxonomic resolution of interest, and calls for the development of more conservative classification methods that are fit for purpose. Thus, with 16S rRNA gene datasets, one must be cautious regarding the detection of taxonomic groups of public health interest (e.g., culture independent identification of foodborne pathogens or taxa associated with a given phenotype).