You are here:
Comparison of Enterococcus qPCR analysis results from fresh and marine water samples on two real-time instruments - poster
Citation:
Sivaganesan, Mano, M. Varma, AND Rich Haugland. Comparison of Enterococcus qPCR analysis results from fresh and marine water samples on two real-time instruments - poster. Presented at ISME14, International Society for Microbial Ecology, Copenhagen, DENMARK, August 19 - 24, 2012.
Impact/Purpose:
To inform the scientific community about the comparison of water samples on two real-time instruments.
Description:
The U.S. Environmental Protection Agency (EPA) will be recommending a quantitative polymerase chain reaction (qPCR) method targeting Enterococcus spp. as an option for monitoring recreational beach water quality. A practical consideration for widespread implementation of this or any other qPCR method is whether the results are comparable on different types of real time PCR instruments. In this study we compared quantitative estimates of enterococci determined by the qPCR method from cycle threshold (Ct) measurements obtained on an Applied Biosystems StepOnePlusTM instrument and on a Cepheid Smart Cycler®. Analyses were performed simultaneously by the two instruments on DNA extracts from multiple, replicate filter retentates of 12 marine and 12 freshwater samples from diverse locations. Three variations of a comparative Ct model, which is used in the Enterococcus qPCR method to estimate relative Enterococcus densities in water samples, were examined in the comparisons. These variations differed in their sources of calibration data used in the model. In addition, both a traditional statistical modeling approach utilizing analysis of variance (ANOVA) and a Bayesian statistical modeling approach that accounts for the influences of calibration and test sample data uncertainty were examined in the instrument comparisons. The traditional analysis approach indicated no significant differences (p>0.05) between the mean density estimates for all samples from the two instruments in two of the three model variations. The Bayesian approach indicated that the 95% Bayesian credible intervals of log10 density estimates for all samples from the two instruments overlapped in all three models, however, the uncertainty of the estimates from the two instruments varied depending on the model. These results support the interchangeable use of the two instruments in the method but also illustrate the value of using both traditional and Bayesian analysis approaches in comparison st