Statistical inference from the fraction labelled mitoses curve

Abstract

SUMMARY: The fraction labelled mitoses curve is widely used to study the premitotic phase of the cell cycle in proliferating cell populations. The population is here assumed to be in a steady state of exponential growth. The theoretical curve is derived in an explicit computational form, allowing for correlations between phases. This gives the likelihood function explicitly, and classical maximum likelihood methods of fitting the theoretical curve to experimental data can work very well. The curve may often be approximated by assuming a joint normal distribution of phase times, and computation of the curve and of the maximum likelihood estimates then becomes straightforward. The mitotic index may sometimes be used to give supplementary information about the mean mitotic time. Weak inferences about the presence or absence of negative correlations between certain phases are possible. The methods are illustrated with three examples of experimental data. © 1970 Oxford University Press.