Optimal design for model discrimination using the F-test with non-linear biochemical models. Criteria for choosing the number and spacing of experimental points