I have a set of samples of different time points, and I'm trying to measure gene expression in treated group compared to control. I'm measuring IL-6 mRNA expression and using the 2-DDCt method. Expression of IL-6 mRNA is almost to non detectable prior to treatment, so therefore I calculated the 2-DDCt as below :
The DCt treated = Ct target gene treated - Ct housekeeping gene of control at each time point
The DCt control = Ct target gene of control - Ct housekeeping gene of control at each time point
The DDCt = DCt treated at particular timepoint - mean of control at particular timepoint
I then use DDCt to calculate the 2-DDCt.
I'm having trouble how to statistically analyse the data - the steps required. I have very little background in statistics, I do use GraphPad but my understanding is only limited to 2-way ANOVA and post-hoc t-test and one-way ANOVA. With the data I have, I'm looking at the difference of expression/folds across time points. Can someone help me the steps for stats using GraphPad?
The data are your dct values. There is no need to manually calculate and ddct values. This will be part of the further analysis anyway.
The experimental design comprises two factors: group and time. If "time" is modeled as a continuous variable, an ANCOVA will be appropriate, if "time" is modeled as a categorical variable ("factor") , then a two-factorial ANOVA is an appropriate choice.
Since both, ANCOVA and two-factorial ANOVA (and one.way-ANVOA and t-tests) are all just parts of special cases of "(general) linear models" I suggest that you get a bit into this topic. Wikipedia is a good start. Googeling for "linear model" will offer you a lot of further sources.
Thank Jochen
"The data are your dct values. There is no need to manually calculate and ddct values. This will be part of the further analysis anyway."
- Do you mean the statistics is calculated using Dct values THEN followed by ddct conversion? Hope you could help clarify that part?
The standard for linear models (such as t-test, ANOVA, ANCOVA and so on) is a "treatment coding" for the levels of categorical predictors. That means that the coefficients are calculated as the differences of the mean value for any level (usually the treatments) and the "reference level" (usually the "untreated" or "control"). When your data are dct-values, then these differences (that are calculated by the linear model) are ddct-values. That's all. If you would provide ddct values, then the differences of such ddct values would be analyzed (dddct values, so to say) what makes little sense.
I'd suggest to read some basics about linear models. The Wikepedia is a good start.
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis