I am measuring degradation of RNA over time with a control and a treatment. I have % RNA remaining (relative to time 0 for control and treatment) and 5 time points. I would like to test if the % remaining is different for control vs. treatment at each time point. I think that I need to correct for multiple comparisons, but unsure of exactly what test to use. I think I need to first do some kind of anova and then a post hoc test such as a t test with a Holm-Bonferroni correction, but this situation looked a bit different from most of the examples I saw where they are testing multiple comparisons within one time point. Thanks


1 Answer 1


You probably shouldn't test for a statistical difference between the control and treatment time-courses at each time point. Are you thinking of hypotheses such as "there is a difference at time point 1 but not time point 3"? No. Look at the experiment as providing estimates of the degradation rate.

The first thing (always) in data analysis is to inspect the data graphically. My expectation is that RNA is pretty unstable and should degrade with an approximately exponential time-course, but you should know more about it and your data have the last say. Plot the amount remaining at the time points using a logarithmic amount scale. You might find a linear decline that is very easily characterised. You might even find that the difference is substantial enough that the graph is sufficiently persuasive without further analysis.

If you do want a more statistical analysis then I suggest fitting a model to the data to determine time-constants. The model to use depend on the form of the data and details of your analytic objectives, but I suggest you look at this question about linear and non-linear regression as a starting point.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.