Abstract
The botanical genus Digitalis is equal parts colorful, toxic, and medicinal, and its bioactive compounds have a long history of therapeutic use. However, with an extremely narrow therapeutic range, even trace amounts of Digitalis can cause adverse effects. Using chemical methods, the United States Food and Drug Administration traced a 1997 case of Digitalis toxicity to a shipment of Plantago (a common ingredient in dietary supplements marketed to improve digestion) contaminated with Digitalis lanata. With increased accessibility to next generation sequencing technology, here we ask whether this case could have been cracked rapidly using shallow genome sequencing strategies (e.g., genome skims). Using a modified implementation of the Site Identification from Short Read Sequences (SISRS) bioinformatics pipeline with whole-genome sequence data, we generated over 2 M genus-level single nucleotide polymorphisms in addition to species-informative single nucleotide polymorphisms. We simulated dietary supplement contamination by spiking low quantities (0-10%) of Digitalis whole-genome sequence data into a background of commonly used ingredients in products marketed for "digestive cleansing" and reliably detected Digitalis at the genus level while also discriminating between Digitalis species. This work serves as a roadmap for the development of novel DNA-based assays to quickly and reliably detect the presence of toxic species such as Digitalis in food products or dietary supplements using genomic methods and highlights the power of harnessing the entire genome to identify botanical species.