Herman Petr, Lee J Ching
Faculty of Mathematics and Physics, Institute of Physics, Charles University, Prague, Czech Republic.
Methods Mol Biol. 2012;796:399-421. doi: 10.1007/978-1-61779-334-9_22.
In this chapter, we demonstrate the advantage of the simultaneous multicurve nonlinear least-squares analysis over that of the conventional single-curve analysis. Fitting results are subjected to thorough Monte Carlo analysis for rigorous assessment of confidence intervals and parameter correlations. The comparison is performed on a practical example of simulated steady-state reaction kinetics complemented with isothermal calorimetry (ITC) data resembling allosteric behavior of rabbit muscle pyruvate kinase (RMPK). Global analysis improves accuracy and confidence limits of model parameters. Cross-correlation between parameters is also reduced with accompanying enhancement of the model-testing power. This becomes especially important for validation of models with "difficult" highly cross-correlated parameters. We show how proper experimental design and critical evaluation of data can improve the chance of differentiating models.
在本章中，我们展示了同时进行多曲线非线性最小二乘法分析相对于传统单曲线分析的优势。对拟合结果进行全面的蒙特卡罗分析，以严格评估置信区间和参数相关性。在一个模拟稳态反应动力学的实际例子上进行比较，该例子补充了类似于兔肌肉丙酮酸激酶（RMPK）变构行为的等温滴定量热法（ITC）数据。全局分析提高了模型参数的准确性和置信限。随着模型测试能力的增强，参数之间的互相关性也降低了。这对于验证具有“困难”的高度互相关参数的模型尤为重要。我们展示了适当的实验设计和对数据的批判性评估如何能够提高区分模型的机会。
