Automation parameter fitting

The refinement is automated by defining a goodness of the fit (GOT) parameter and using numerical optimization routines to do the search in a computer. One of the most useful GOF s for direct comparison between experimental and theoretical intensities is the... 

Although the MM equation is a powerful kinetic form to which the vast majority of enzyme kinetics has been fitted, one should not forget the assumptions and limitations of the model. As a basic example, feedback inhibition, whereby the product of the reaction inhibits the enzyme-substrate cooperativity, multiple-substrate reactions, allosteric modifications, and other deviations from the reaction scheme in equation (1) are treated only adequately by the MM formalism under certain experimental conditions. In other words, enzyme kinetics are often bent to conform to the MM formalism for the sake of obtaining a set of parameters easily recognizable by most biochemists. The expUcit mathematical and experimental treatment of reaction mechanisms more complex than that shown in equation (1) is highly involved, although a mathematical automated kinetic equation derivation framework for an arbitrary mechanism has been described in the past (e.g., ref. 6). 

An automated parameterization may be difficult to set up. But when this has been accomplished, the process is substantially faster than the manual method, and much larger bodies of data can be fitted simultaneously. The main drawback of the automated scheme is that errors may remain undetected more easily than in manual parameterization. Automated parameterization therefore requires substantial validation to identify outliers in the data set and deficiencies in the force field. Statistical tools should be used to verify that each parameter is well defined by the chosen set of reference data, and any iU-fitting data points can be rationalized on sound physical grounds. 

The methodologies used for fitting calibration curves depend on whether they are linear or nonlinear. Model fitting basically consists of finding values of the model parameters that minimize the deviation between the fitted curve and the observed data (i.e., to get the curve to fit the data as perfectly as possible). For linear models, estimates of the parameters such as the intercept and slope can be derived analytically. However, this is not possible for most nonlinear models. Estimation of the parameters in most nonlinear models requires computer-intensive numerical optimization techniques that require the input of starting values by the user (or by an automated program), and the final estimates of the model parameters are determined based on numerically optimizing the closeness of the fitted calibration curve to the observed measurements. Fortunately, this is now automated and available in most user-friendly software. 

Now to our subject. Modern software programs offer many options for the manipulation of the chromatogram, e.g. automated baseline subtraction, programmed change of integration parameter, auto-zero functions and auto scale, automated calibration for non-linear signal-to-concentration ratio, etc. It is worth while knowing these options and how to use them as they fit. 

---