Explore a new proof-of-concept study that combines mitochondrial DNA hybridization capture with a k-mer-based identification pipeline to achieve species-level detection of arthropod contaminants in complex food matrices with greater sensitivity than traditional microscopy.
The contamination challenge: where microscopy reaches its limits
Regulatory oversight of arthropod contamination in food has long depended on physical extraction followed by microscopic examination. Analysts classify fragments to the most specific taxonomic level possible, a process that is labor-intensive, requires deep morphological expertise, and frequently cannot resolve fragments beyond order or family level. The problem is compounded in processed foods, where grinding, milling, and heat treatment pulverize arthropod material into fragments that lack the diagnostic features needed for species-level identification.
As global food supply chains diversify and surveillance increasingly emphasizes source attribution, the limitations of microscopy are becoming more consequential. Distinguishing a stored-product pest from a field pest, or confirming the presence of a known foodborne illness vector, requires a level of taxonomic resolution that damaged fragments often cannot provide. A more sensitive and taxonomically informative approach is needed.
The study: a mito-metagenomic approach to food safety genomics
Researchers at the U.S. Food and Drug Administration (FDA) developed a targeted metagenomics workflow designed to address these limitations. Published in Food Control, the study¹ describes a method combining mitochondrial DNA hybridization capture with the Mitok-mer pipeline, a k-mer-based taxonomic assignment approach for eukaryotes, to achieve species-level identification of arthropod contaminants in food-associated matrices.
The workflow is built on MitochonTrakr, the FDA’s curated mitogenome reference collection, which grew from 13,036 eukaryotic mitogenomes in 2020 to 65,443 in 2023. Short mitochondrial k-mers derived from this database are used both to select informative targets for bait design and to classify sequencing reads at the species or near-species level. The approach exploits a key biological advantage: mitochondrial genomes are present in high copy number relative to nuclear DNA, which improves signal-to-noise and provides higher phylogenetic resolution than short barcodes such as COX1 alone.
Hybridization capture: filtering signal from noise
The central challenge in food safety metagenomics is not sequencing capacity. When arthropod DNA represents a fraction of total DNA in a wheat flour or cornmeal sample, even mitochondrial reads can be sparse in untargeted libraries, limiting sensitivity and increasing sequencing costs.
To overcome this, the FDA team designed a custom myBaits target capture kit through Daicel Arbor Biosciences. Two panel versions were developed. The SnackAttack-BugTrakr v1 panel contains approximately 280,000 80-mer biotinylated RNA baits targeting 2,536 mitogenomes representing 1,988 arthropod species across 42 orders. The expanded v2 panel contains approximately 480,000 110-mer baits targeting 7,205 mitogenomes from 5,522 species across 49 orders, with increased redundancy for high-priority regulatory taxa like Diptera (including Chrysomya, Musca, and Lucilia). Both panels were designed by mapping species-informative 30-mer k-mers back to their source mitogenomes, merging overlapping hits into short target loci, and providing these curated loci to Daicel Arbor Biosciences for panel optimization and bait synthesis.
In practice, pooled metagenomic libraries are incubated with the biotinylated RNA baits, which hybridize to arthropod mitochondrial sequences in the mixture. Bait-bound fragments are then captured on streptavidin-coated magnetic beads and washed to remove non-target DNA before amplification and sequencing. The food matrix DNA, which would otherwise dominate the sequencing output, is left behind.
Sensitivity and specificity: performance in mock communities
The workflow was validated using mock communities: purified insect genomic DNA spiked into wheat flour or cornmeal backgrounds at defined concentrations. In 30-species wheat-flour mixtures, hybridization capture increased the mean on-target arthropod signal from 80.9% to 98.4%, while reducing wheat-derived signal by 1,400 to 2,000-fold. Species-level relative abundances remained broadly consistent between enriched and unenriched libraries (Pearson r = 0.9), confirming that capture concentrated arthropod signal without distorting community composition.
The gains were more pronounced in cornmeal matrices, where arthropod DNA was present within a larger background. In single-species cornmeal spike-ins, the target insect increased from a range of 4.7 to 53.2% relative abundance in shotgun libraries to 87.9 to 99.6% after bait capture, an average 6.8-fold enrichment. Non-target signal, attributed by the cornmeal matrix and non-target arthropod DNA, was depleted by 8 to 111-fold. In five-species cornmeal mixtures, all five target taxa reached 99% relative abundance after capture, and the closely related beetles Tribolium castaneum and T. confusum were resolved to species level even at the lowest insect load tested. The largest reduction in background noise occurred at the lowest insect input, where combined non-target signal dropped 53-fold, demonstrating that hybridization capture is particularly effective precisely when sensitivity matters most.
A preliminary assessment of two naturally contaminated food products, sandwich cookies and dried fish maw, provided an initial bridge to real-world application. In the sandwich cookie sample, targeted capture increased combined arthropod DNA relative abundance from 0.11% to 63.09% while reducing wheat-derived DNA abundance from 59.85% to 16.31%. These results show that taxonomically informative arthropod DNA can be recovered from real-world processed products.
Actionable data: what genomic identification enables
The practical value of this workflow extends well beyond the detection of arthropod presence. When a regulatory investigation requires determining whether a contaminant is a stored-product pest, an agricultural field species, or a potential vector of foodborne illness, species-level resolution matters. Microscopic observation of arthropod fragments cannot reliably provide that resolution. Genomic identification can.
The authors note that targeted metagenomics can reduce reliance on specialized fragment identification and expert morphological taxonomy, areas in which long-term capacity may be difficult to sustain as experienced analysts retire. The approach is well-suited for a staged filth-analysis workflow: targeted capture could help prioritize samples containing arthropod taxa of greatest regulatory interest, with microscopy retained for cases where developmental stage, specimen condition, or fragment-based metrics are needed for enforcement.
The custom bait design process also provides flexibility for future applications. Panels can be scaled to cover additional arthropod taxa, geographic populations, or commodity-specific threat profiles, and the v2 panel’s expanded mitogenome redundancy is designed to improve robustness to probe-target mismatch in samples containing unanticipated diversity.
Both panels (v1: D10740MTIN2 and v2: D10784ISMT3) are available to researchers.
Design a custom panel for your metagenomic research
The FDA study demonstrates that hybridization capture is a practical, high-performance tool for targeted metagenomics in complex food matrices. Whether your research involves arthropod detection in processed foods, non-model organism genomics, or other applications where rare targets are buried in high-background DNA, custom myBaits panels can be tailored to your specific targets and taxonomic scope.
Explore myBaits Custom Target Capture design services to learn how targeted enrichment can improve your sequencing outcomes.
References
- Pava-Ripoll, M., et al. Targeted metagenomics enables species-level detection of arthropod contaminants in food. (2026) Food Control 188, 112217. https://doi.org/10.1016/j.foodcont.2026.112217
Bluesky
X
Linkedin
Place order
Quote request
Support
Cart