Abstract
Polymerase chain reaction (PCR) is a laboratory procedure to amplify and simultaneously quantify targeted DNA molecules, and then detect the product of the reaction at the end of all the amplification cycles. A more modern technique, real-time PCR, also known as quantitative PCR (qPCR), detects the product after each cycle of the progressing reaction by applying a specific fluorescence technique. The quantitative methods currently used to analyze qPCR data result in varying levels of estimation quality. This study compares the accuracy and precision of the estimation achieved by eight different models when applied to the same qPCR dataset. Also, the study evaluates a newly introduced data preprocessing approach, the taking-the-difference approach, and compares it to the currently used approach of subtracting the background fluorescence. The taking-the-difference method subtracts the fluorescence in the former cycle from that in the latter cycle to avoid estimating the background fluorescence. The results obtained from the eight models show that taking-the-difference is a better way to preprocess qPCR data compared to the original approach because of a reduction in the background estimation error. The results also show that weighted models are better than non-weighted models, and that the precision of the estimation achieved by the mixed models is slightly better than that achieved by the linear regression models.