Statistical comparison of ligand-binding kinetics

We describe the application of generalized linear model methodology to the problem of testing differences among ligand-receptor interactions, and show that the method is analogous to weighted least squares regression methodology and F tests familiar to many investigators. The method accommodates incomplete block designs so that one can obtain kinetic parameter estimates directly comparable among samples analysed on incompletely overlapping sets of experimental runs. We demonstrate the method with data that compare saturation kinetics for a single penicillin-binding protein in isogenic ampicillin susceptible and resistant bacteria.