Comparing time curves for two groups in research.

#1
Hello, I am creating time curves (experimentally) for two groups of mice (f(t) and y(t)): number of particular cells (in murine blood, per ml of blood) vs time after single intravenous injection of one drug (which is supposed to induce proliferation of the cells under study). I chose to start with 8 mice per group: 8 wild type and 8 knock out mice (for LDLR gene). At time "0" there is no difference between the groups in the number of studied cells. At later time points I see the difference between the two groups of mice which is significant (by Mann-Whitney test (MW) at particular time points) and increases with time (knock out animals are not so sensitive to the drug administration as wild type).

What is the standard way (for publication) to compare these two groups of mice? How do I determine whether the two groups differ in the ability to induce proliferation of studied cells. My first guess is to calculate the area under the curve for each mice and compare the AUCs for two groups by MW test. But I want to know for sure whether this is a standard way to analyze this kind of data. It is possible that in spite of the similar AUC values between the groups the time curves differ. Is there a standard way to account for this difference? Should I construct

z=∫sqrt([f(t)-averagef(t)]2)dt

for every wild type mice time curve (f(t)), create a 95% confidence interval for z by MW test and check whether z=∫sqrt([averagey(t)-averagef(t)]2)dt lies in this confidence interval?

Can I do it with only 3-5 mice per group? I know minimal number of data points per group in MW test is 8, but I also know that many people in the literature use only 4-6 mice per group..

Can I do it in Graphpad Prism or SPSS?
Thank you!
 
#2
Your measurements are actually discrete. You are not examining the mice every millisecond. If the number of studied moments of time is not large, you can try Repeated Measures ANOVA, with Group being a between-subject effect.

If the number of studied moments of time is large, you can estimate a time series model for each group and compare the relevant coefficients (e.g. long-run mean, trend etc) between the groups using a likelihood ratio test (or a similar test).
 
#4
Yes, but you should run diagnostic tests (like the sphericity test) to see if the assumptions of Repeated Measures ANOVA hold.... Not a case for time series if you have only 10 time points.