Description:
<jats:p>A strategy suitable for computer programming is presented which permits the estimation of kinetic parameters and of their range of statistical variation from initial‐rate measurements or from binding data. It can be used for steady‐state rate equations, rapid‐equilibrium mechanisms, binding kinetics, descriptive models such as Adair's or Hill's equations, and for the study of more complicated allosteric models. The desired kinetic equation is addressed from a model catalogue.</jats:p><jats:p>The strategy consists of four main steps: (1) formulation of a goodness‐of‐fit criterion using the concept of the distance between observed and predicted behaviour of the model, avoiding assumptions about the statistical structure; (2) localization of the best fit, <jats:italic>i.e.</jats:italic> of the parameter point of minimum distance from the data by a search algorithm derived from Marquardt's method; (3) exclusion of discrepancies between model and data by testing the randomness of the vector of residual deviations and the structure of the information matrix; (4) exploration of the parameter variability based on a linear approximation of the fit criterion.</jats:p><jats:p>Input to the programme are ligand concentrations and measured‐response data, a code number of the desired model, and a guessed starting point of the parameters which, in contrast to older methods, need not be precise. The output is either a set of parameter values with a respective region of variation, supplemented by tables and graphs of the fit obtained, or the rejection of the data with comments on the reason of failure. The possible reasons include statistical fluctuation or systematic trends in the data, lack of required information (poor experimental design), and inappropriate model (non‐unique or inconsistent kinetic model).</jats:p><jats:p>Tests on isosteric and allosteric kinetic data lead to the conclusion that the convergence of the strategy is very good when the fit surface is well defined (good data and correct model).</jats:p>