I have gene expression data collected from different animals 1, 3, 9, and 24 hours after drug administration.
I also have vehicle-injected animal values 1, 3, 9, 24 hours later.
Plotting these data reveals a relatively obvious pattern between groups, but I would like to collapse the time-frame into a single bar graph.
If values were taken from the same animal 1, 3, 9, and 24 hours after injection, I could simply use area under the curve. However, in this case, different animals were used at each time point. Is there a valid statistical technique/test I could use?
Thanks for any help.
Hi Anuja
Thanks for your answer. I have SPSS and Prism, so can input data into either. I will try the log transform method you suggest and see how that works. Thanks!
Hi Jeremy.
Why do you want to collapse the time frame? As I understand you have a 2x2 factorial design (1. drug vs placebo, 2. time frame). You are mainly interested in the treatment effect. Simply use an ANOVA with gene expression ~ factor(treatment) + factor (time) + factor(treatment)*factor(time). The main effect for "treatment" tells you if there is a treatment effect at 1 hour. The main effect(s) of "time" tell you if the gene expression in general changes over time in comparison to 1 hour. The interaction tells you if the treatment effect changes over time. If you are interested in the marginal effect of "treatment" you simply fix the time effect at a certain time value e.g. mean time value.
Log transformation (ln) would be a good place to start. AUC would be a good way to characterize the change in gene expression - Unless controls are all 0. If the data are normally distributed, Stefan's ANOVA would be interesting as well.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques