This document describes the development of a high-throughput universal digital high-resolution melt (U-dHRM) platform for rapid bacterial identification. The platform aims to address limitations of current diagnostic methods by enabling broad-based, sensitive, rapid and polymicrobial detection of bacteria from small sample volumes. Key aspects of the platform include universal 16S amplification from bacterial DNA, partitioning of the reaction into individual wells of a digital PCR chip for sensitive detection, and high-resolution melt analysis of the amplified products to identify bacterial strains. Current work is focused on transitioning to digital droplet PCR and developing a portable system to enable clinical applications of this rapid bacterial identification approach.
3. Motivation – Sepsis
>750,000 cases in the US each year
26% - 50% mortality rate
One of top 10 causes of infant mortality
Tricky to diagnose
Non-specific symptoms
Unhelpful diagnostic protocols
Overwhelming and life-threatening inflammatory response to
a bloodstream infection.
4. Motivation – Blood Culture
Current gold standard
Can take several days or more for a result
Difficult to determine bacterial strains
Broad-spectrum antibiotics in the interim
Negative results don’t necessarily affect
diagnosis!
5. Motivation – Diagnostic Criteria
Broad-based – detect all bacteria in sample
Sensitive – small sample volumes and 1 – 2000 CFU/mL
Rapid – targeted antibiotic treatment based on specific strains
Polymicrobial – precise composition of heterogeneous samples
6. Background – Molecular Diagnostics
DNA Microarrays
Quantitative Real-time PCR
Rapid, no need for culture
Lacks broad-based detection
Hybridization inaccuracies
7. Background – 16S PCR
16S rRNA gene
Present in all bacteria broad-based
Strain-specific hypervariable regions
Prone to false positives and negatives from small volumes (1-5 uL)
Still need to sequence amplicon
8. Background – High Resolution Melt
Rapid, inexpensive post-PCR
sequencing alternative
Non-specific intercalating dye
Solely dependent on sequence
Constrained to homogeneous samples
9. Background – dPCR
Scalable dynamic range and cost-effective
DNA occupancy follows a Poisson distribution
10 - 100 wells to DNA molecules
96 wells up to 9 molecules
20,000 wells up to 2,000 molecules
Neonate blood contains 1-2,000 CFU/mL
Partitions sample into individual
reaction wells polymicrobial
& sensitive
10. Universal Digital HRM (U-dHRM)
16S PCR broad-based, rapid
Digital PCR / Microfluidics sensitive, polymicrobial
High Resolution Melt cost-effective
13. U-dHRM – One-versus-one Support
Vector Machine (OVO SVM)
Supervised – learns from labeled training data (support vectors)
Binary – determines between 2 classifications
L. monocytogenes
S. pneumoniae
Unknown Curves Training Data Classified Curves
14. Current Work – Brief
Transition to digital droplet PCR
Scalable dynamic range
High-throughput
Portability
Image entire chip
Remove microscope
15. Summary
U-dHRM Platform
Broad-based – universal 16S amplification
Polymicrobial – highly accurate profiling of heterogeneous samples
Sensitive – dPCR chip
Rapid – DNA amplification strategies and HRM
Current Work
Provides high-throughput format
Improves portability for clinical applications
16. Acknowledgments
Dr. Colleen Chute-Ricker
Dr. Julietta Jupe
Daniel Ortiz
Tyler Goshia
Anthony Han
Hannah Mack
Nick Teuthorn
Many thanks to all the members of the Fraley Lab
Many thanks to our collaborators
Dr. Shelley Lawrence – Assistant Professor, Pediatrics, UCSD Medicine
Dr. Hannah Carter – Assistant Professor, Medical Genetics, UCSD Medicine
Brian Tsui – Carter Lab
Mridu Sinha – Coleman Lab
Many thanks to our sponsors