A Schema for Digitized Surface Swab Site Metadata in Open-Source DNA Sequence Databases

**Background:** Modern surveillance of foodborne pathogens relies on large, open-source DNA sequence databases such as NCBI's Pathogen Detection Database. Environmental monitoring through surface swabbing in built environments is critical for identifying point sources of outbreaks. However, metadata describing swab site locations are typically collected as unstructured free-text responses, which are often incomplete, inconsistent, and difficult to analyze at scale. This lack of standardization hinders machine-readability, slows epidemiological investigations, and complicates integration across digital networks. The study aimed to develop a standardized schema for swab site location metadata using existing ontologies and the OBO Foundry Principles.

**Methods:** The researchers studied 1,498 swab site descriptions generated during routine food safety surveillance by the U.S. FDA. Records came from 9 facilities (one dairy, four produce, one seafood, one mixed facility) across 22 collection dates via convenience sampling. All facility-specific information was anonymized. Free-text responses were analyzed using the tm v0.7-8 text mining tool in R version 4.1.2. Two researchers independently performed emergent coding to categorize components into informational facets, achieving >95% agreement. Each unique term was queried in Ontobee; terms not found were proposed as new terms to existing OBO ontologies using ROBOT templates. Twenty-one term additions were requested across three ontologies (15 to ENVO, 1 to PATO, 5 to NCIT), all of which were fulfilled. A rarefaction curve was generated using the vegan package v2.5-7. A chi-square test of association with Bonferroni correction was performed in R v4.1.2 to identify structures with significantly higher *L. monocytogenes* incidence.

**Key Results:** Analysis of free-text responses identified 338 unique terms across 1,498 descriptions. The majority (n=253) described the structure and subpart being swabbed. Five informational facets were identified: (i) structure being swabbed, (ii) subpart of the structure, (iii) material of the swab surface, (iv) condition of the surface, and (v) orientation of the swab site. The structure was defined in 99.6% of metadata, but other facets were addressed far less frequently. The rarefaction curve revealed high term diversity, indicating a need for controlled vocabulary. In the use case analysis of *L. monocytogenes* detection, wheels (28 out of 128) and brooms (8 out of 11) had significantly higher proportions of positive samples (P < 0.05) compared to other commonly selected locations. The schema was implemented in the One Health Enteric Package, an expanded metadata suite for genomic surveillance of enteric pathogens developed by the U.S. interagency GenFS working group.

**Clinical Implications:** Standardized, machine-readable swab site metadata can significantly expedite outbreak investigations by reducing the interpretative labor needed to extract actionable information from free-text responses. The schema enables rapid quantitative evaluation of surveillance efforts, allowing agencies to identify high-risk locations with increasing specificity. Implementation through digital tools with selectable drop-down menus can assist data collectors with real-time data capture. The framework supports the FAIR principles (Findability, Accessibility, Interoperability, Reusability) and is compatible with FoodOn and GenEpio, facilitating integration of genomic epidemiological data with food product data. International alignment on contextualizing metadata is increasingly important for global food supply chains. The One Health Enteric Metadata Package, hosted at NCBI BioSample, now includes this framework for samples collected during facility inspections, implementing the schema for national and international genomic surveillance of foodborne pathogens.