A specific statistical model and algorithm related to the detection of Mollicutes in contaminated biological samples by Real-Time Transcription Mediated Amplification.

摘要

Among all nucleic acid amplification technologies, Real-Time Transcription Mediated Amplification (Real-Time TMA) is an isothermal method that can amplify RNA targets a billion-fold in less than one hour's time. By using this method, a new assay was developed for detecting the presence of Mollicutes in mammalian cell cultures and biologics. Production of amplicons is monitored in real time by measuring continuously a fluorescence signal during the reaction. The shape of this signal curve is a sigmoid, where an initial baseline phase precedes an exponential phase ending with a maximum followed by a linear decreasing phase. The aim of this study was to develop a curve-analysis tool to unambiguously assign a Mollicutes positive or negative status to a biological sample. In this context, a statistical analysis of the data combined with the selection of the best predictors allowed the development of an algorithm which removes erroneous data and applies the best predictors to generate the Mollicutes status prediction. Our results demonstrate that this algorithm assigns positiveegative/invalid status coming from Real-Time TMA fluorescence signal analysis with a confidence (true predictions) in the results close to 100%.