Cell Culture Development, Global Biological Development, Bayer HealthCare, Berkeley, CA 94710, USA.
Biotechnol Prog. 2011 Sep-Oct;27(5):1407-14. doi: 10.1002/btpr.666. Epub 2011 Jul 15.
We present an alternate approach for analyzing data from real-time reverse transcription polymerase chain reaction (qRT-PCR) experiments by fitting individual fluorescence vs. cycle number (F vs. C) curves to the logistic growth equation. The best fit parameters determined by nonlinear least squares were used to compute the second derivative of the logistic equation and the cycle threshold, C(t), was determined from the maximum value of the second derivative. This C(t) value was subsequently used to determine ΔΔC(t) and the amplification efficiency, E(n), thereby completing the analysis on a qRT-PCR data set. The robustness of the logistic approach was verified by testing ~600 F vs. C curves using both new and previously published data sets. In most cases, comparisons were made between the logistic estimates and those from the standard curve and comparative C(t) methods. Deviations between the logistic and standard curve method ranged between 3-10% for C(t) estimates, 2-10% for ΔΔC(t) estimates, and 1-11% for E(n) estimates. The correlations between C(t) estimates from the logistic and standard curve methods were very high, often >0.95. When compared with five other established methods of qRT-PCR data analysis to predict initial concentrations of two genes encompassing a total of 500 F vs. C curves, the logistic estimates were of comparable accuracy. This reliable performance of the logistic approach comes without the need to construct standard curves which can be a laborious undertaking. Also, no a priori assumptions for E(n) are necessary while some other methods assume equal E(n) values for the reference and target genes, an assumption that is not universally valid. In addition, by accurately describing the data in the plateau region of the F vs. C curve, the logistic method overcomes the limitations of the sigmoidal curve fitting method. The streamlined nature of the logistic approach makes it ideal for complete automation on a variety of computing environments thereby completely eliminating user bias. The simplicity, robustness, and ease of computer implementation of the logistic approach should make it an attractive alternative for rapidly analyzing qRT-PCR data.
我们提出了一种分析实时逆转录聚合酶链反应（qRT-PCR）实验数据的替代方法，即将单个荧光与循环数（F 与 C）曲线拟合到逻辑增长方程。通过非线性最小二乘法确定的最佳拟合参数用于计算逻辑方程的二阶导数，并且通过二阶导数的最大值确定循环阈值（C(t)）。然后使用此 C(t)值来确定 ΔΔC(t)和扩增效率（E(n)），从而完成 qRT-PCR 数据集的分析。通过使用新的和以前发布的数据集测试了大约 600 个 F 与 C 曲线，验证了逻辑方法的稳健性。在大多数情况下，将逻辑估计值与标准曲线和比较 C(t)方法的估计值进行比较。逻辑估计值与标准曲线方法之间的偏差范围为 C(t)估计值的 3-10%，ΔΔC(t)估计值的 2-10%，E(n)估计值的 1-11%。逻辑方法和标准曲线方法的 C(t)估计值之间的相关性非常高，通常 >0.95。当与另外五种已建立的 qRT-PCR 数据分析方法比较以预测涵盖总共 500 个 F 与 C 曲线的两个基因的初始浓度时，逻辑估计值具有相当的准确性。逻辑方法的这种可靠性能无需构建标准曲线，而标准曲线可能是一项繁琐的工作。此外，不需要对 E(n)进行先验假设，而其他一些方法则假设参考基因和靶基因的 E(n)值相等，这一假设并非普遍适用。此外，通过准确描述 F 与 C 曲线的平台区域中的数据，逻辑方法克服了 S 形曲线拟合方法的局限性。逻辑方法的流线型特性使其非常适合在各种计算环境中完全自动化，从而完全消除用户偏见。逻辑方法的简单性、稳健性和易于计算机实现使其成为快速分析 qRT-PCR 数据的有吸引力的替代方法。
