Experimental Designs for the Distribution-Free Analysis of Enzyme Kinetic Data

Abstract

The statistical analysis of experimental data always involves making certain assumptions about the distribution of experimental errors. Methods based on the least-squares principle are appropriate when these errors follow a normal distribution but several studies have shown that, for enzyme kinetic measurements, such a distribution may not be common. Under these circumstances, distribution-free methods of analysis should be employed. In this paper a distribution-free method, which can be applied to a wide variety of data analysis problems, is proposed. The method is based on a special type of experimental design in which replicate measurements are made under as many experimental conditions as there are parameters to be estimated. This design offers great simplicity in the execution of the experiment and in the analysis of the results. In addition, the design can be made optimal in the sense that the overall variance of the parameter estimates is minimized. Formulae are given for choosing the optimal designs for selected cases and a worked example of the design and analysis of a competitive inhibition experiment is presented.