UC San Diego

High Resolution Melt (HRM) analysis allows for the classification of DNA sequences based on the way the sequences melt when exposed to heat in the presence of an intercalating dye. While HRM has been shown to be effective in broad-based sequence identification, it has remained limited in its sensitivity and is unable to provide information regarding absolute quantification of samples. Here, HRM is combined with digital PCR by isolating individual templates into separate wells and melting the end-point product, producing a highly sensitive technique capable of sequence identification and absolute quantification in multiplexed samples. By decreasing the reaction volume by more than 99% and increasing the number of reactions to 20,000, this novel platform offers new detection capabilities not previously seen in standard qPCR HRM. By optimizing temperature resolution, reagent concentrations, sampling rate, and microscope exposure settings, the platform’s performance was shown to rival the resolution demonstrated by qPCR HRM. This resolution is of particular importance in the identification of bacteria based on their sequence-specific melt curves. In a proof of concept, by using the universal amplification of the 16s rRNA gene of bacteria, this dHRM platform is demonstrated to be capable of differentiating between two strains of bacteria based on the fingerprint melt curves generated from single-molecule amplification. By training a machine-learning support vector algorithm (SVM) coupled with an automated image analysis code, the platform successfully classified separate strains within a mixed culture of bacteria. These results suggest that this technology may have broad applications for sensitively and rapidly profiling bacterial populations, particularly in clinical cases of infection.