Skip to main content

Evaluation of a next generation sequencing-based approach for infectious disease diagnostics in the National Institutes of Health Clinical Center

Thursday, September 17, 2015 — Poster Session II

12:00 p.m. – 1:30 p.m.
FAES Terrace


  • PP Khil
  • JH Youn
  • AM Zelazny
  • JP Dekker
  • KM Frank


Traditional microbial diagnostics in the clinical lab has relied primarily on microscopy, culture, and targeted amplification by PCR. However, many human pathogens cannot be readily cultured, and the occurrence of infectious syndromes of unknown etiology argues strongly for the existence of as-yet uncharacterized pathogens. Such novel agents (particularly viruses) may not be detected by PCR assays, which require prior-characterized target sequences for their design. An attractive alternative to traditional methods for infectious disease diagnostics is the use of direct next generation sequencing (NGS) of primary specimens. Here we report steps towards the validation and implementation of an integrated NGS-based diagnostic system in the NIH Clinical Center. The logic of this strategy is straightforward and based on a metagenomics approach to the detection of DNAs directly in patient specimens. Implementation, however, is less straightforward. Accurate classification of pathogen-derived reads, subtraction of human reads, strategies for dealing with background DNA contamination, and development of interpretation schemes for clinical reporting are non-trivial. We’ve evaluated our approach using >100 samples, including defined DNA mixes, bacteria spiked into whole blood, clinical specimens with previously characterized pathogens, and pathogen-negative CSF, blood, urine, biopsy, pleural fluid, and peritoneal fluid controls. To improve sequencing read classification we integrated output of two metagenomics pipelines in conjunction with curated and bulk reference databases. We found that non-curated full GenBank dataset usually provides excellent species resolution. Overall, we estimate that the NGS-based approach followed by MiSeq sequencing detects no less than 1 pathogen per 1000 human cells and offers superb specificity.

Category: Microbiology and Infectious Diseases