Ocular Microbiology and Immunology Group
Back to OMIG Main Page

< Previous | 2026 Agenda and Abstracts | Next >

2026 OMIG Abstract

Full-Length 16S Rdna Sequencing with Contamination-Corrected Thresholds Identifies Pathogens in Culture-Negative Infectious Keratitis

Bridget Slomka¹, Vridhi Vinaykiya¹, James Chodosh², and Paulo J. M. Bispo¹

¹Department of Ophthalmology, Mass. Eye and Ear, Harvard Medical School, Boston, Massachusetts; and ²Department of Ophthalmology, University of New Mexico School of Medicine, Albuquerque, New Mexico

Purpose: Targeted 16S rDNA sequencing can reduce time-to-diagnosis in sight-threatening keratitis when compared to bacterial culture. However, this approach is vulnerable to contaminants that confound interpretation. We present two species-specific thresholding approaches to separate pathogen signal from background in low biomass corneal samples.

Methods: Full-length 16S rDNA amplification and Nanopore sequencing were performed on 139 corneal ulcer swabs (culture-positive n=41, culture-negative n=98), and species assignments were performed with Emu. Samples were then categorized as sequencing-positive, sequencing-negative, or inconclusive via manual review of relative abundances using pre-defined thresholds. Species alpha-diversity metrics were compared across sequencing status groups and evaluated as discriminators of sequencing-positive status. To separate contaminant from clinically informative signal, we compared thresholds derived from residual taxa in a subset of culture-verified, sequencing-positive samples and sequencing-negative samples. For the positive-residual approach, the presumptive pathogen was computationally removed from each sample, relative abundances were recalculated, and the residual community was used to derive species-specific background thresholds. For the sequencing-negative approach, relative abundances were used to set contaminant thresholds.

Results: Manual classification of species assignments yielded 57 sequencing-positive, 45 sequencing-negative, and 37 inconclusive classifications. Shannon Entropy, Inverse Simpson Index, and Berger-Parker Index, differed significantly across sequencing status groups (all p < 0.0001, Kruskal-Wallis), with sequencing-positive samples showing markedly lower diversity (Shannon H=1.26 vs 2.19 and 2.25 in negative and inconclusive groups, respectively). The sequencing-negative-derived thresholding approach identified more species as contaminants (n=579) than the positive-residual-derived thresholding approach (n=322) and accurately identified all sequencing-positive samples (n=57/57). Conversely, the positive-residual-derived threshold method identified fewer sequencing-positive samples (n=54/57). The positive-derived thresholds flagged elevated taxa in the negative samples ~6x more often (n=285) than in the sequencing-negative-derived threshold method (n=47).

Conclusions: We demonstrate that species-specific contaminant thresholds derived from sequencing-negative clinical specimens effectively discriminated pathogen signal from background contamination. This approach provided the best overall balance between reducing background signal and preserving expected signal. Differences in species diversity metrics between groups confirmed that contamination community structure encodes potential diagnostic information, independent of pathogen identity. These findings support the use of diversity metrics and data-derived species-specific thresholds as a framework for pathogen detection by 16S rDNA sequencing in low-biomass keratitis specimens.