Source: UNC Health, by Qingyun Liu, University of North Carolina at Chapel Hill, Department of Genetics (October 7, 2026). AI-generated summary by biochip.com, published . Not independently reviewed.
Key takeaways
- UNC and Tulane's CANDI platform detects 15 Mycobacterium species and subspecies using barcoded CRISPR reactions on a microfluidic chip.
- In 125 participants, CANDI achieved 96.0 percent concordance with culture and returned species-level results in approximately 3.5 hours.
- Multicenter clinical validation and treatment-monitoring studies are planned, so broader real-world performance has not yet been established.
Researchers at the University of North Carolina at Chapel Hill and Tulane University have developed a CRISPR-based test that identifies 15 clinically relevant nontuberculous mycobacteria, or NTM, from patient specimens in about 3.5 hours. The system, called CRISPR-Assisted Nanodroplet Differential Identification, or CANDI, combines population-scale genome analysis with a microfluidic chip that runs thousands of tiny, separated reactions at once. In a 125-participant study, it showed 96.0 percent agreement with conventional culture results. Faster species-level identification could help clinicians select care strategies sooner for infections whose treatment can vary sharply by bacterial species.
Why NTM Identification Is Hard
NTM are environmental bacteria in the Mycobacterium genus, distinct from the bacteria that cause tuberculosis. Dozens can cause illness in people, but individual species and subspecies differ in how likely they are to cause disease and how they respond to treatment.
That creates a diagnostic problem. Culture-based workflows can take days to weeks, and a test must distinguish closely related bacteria without missing the natural genetic variation found across real patient populations.
Choosing Targets From 100,000 Genomes
Qingyun Liu's laboratory at UNC worked with Melissa B. Miller, director of the UNC Clinical Microbiology Laboratory, to review seven years of diagnostic records across the UNC Health system. That review helped define the 15 species and subspecies that the initial diagnostic panel needed to cover.
The team then analyzed more than 100,000 mycobacterial genomes. Kelly Eick and Duy Tran evaluated candidate DNA regions, looking for sequences conserved within each target species but distinct enough to separate it from other mycobacteria. Think of this step as finding a spelling pattern shared by nearly every edition of one book but absent from a closely related title.
A Microfluidic Chip for Parallel CRISPR Reactions
The Tulane group, led by Bo Ning and Tony Hu, built the CANDI platform around CRISPR, the programmable molecular system best known for gene editing. Here, CRISPR acts as a highly specific detector: when it recognizes a selected bacterial DNA sequence, it produces a fluorescent signal rather than altering genes.
CANDI's keyhole-shaped microfluidic chip, designed by Hongquan Gou and Long Chen, organizes thousands of miniature CRISPR reactions in parallel. Each compartment receives DNA from the same specimen but carries a unique fluorescent barcode and a different target, allowing the system to check many bacterial identities at once while keeping reactions physically separated.
Performance in Isolates and Patient Specimens
In retrospective testing of archived, culture-derived clinical isolates, CANDI reached 98.20 percent sensitivity and 99.65 percent specificity. Sensitivity measures how often a test detects samples that are truly positive, while specificity measures how often it correctly rules out samples that are negative.
The researchers also tested residual specimens from 125 participants receiving routine care, avoiding extra sample collection. The group included 61 culture-positive and 64 culture-negative individuals. CANDI correctly classified every culture-negative sample, achieved 97.08 percent overall sensitivity and 99.73 percent specificity, and delivered species-level results in roughly 3.5 hours.
Why This Matters
For NTM disease, identifying the organism at species level is not a minor laboratory detail. Different organisms can carry different clinical implications and may require different treatment approaches, making earlier identification potentially useful for clinical decision-making.
Liu and Ning have received a five-year multi-principal investigator R01 grant to expand CANDI to about 30 NTM species and subspecies. The next phase will include multicenter validation, including retrospective and prospective cohorts, and will examine whether changes in bacterial DNA signals can track treatment response, first in mouse models of M. avium and M. abscessus infection and later in patient cohorts.
Topics: Microfluidics · Diagnostics & precision medicine · Infectious diseases
Start here: Microfluidics explained