I have carried out an experiment on cells from the same cell line, grown in two plates and exposed to different conditions (Control vs Treated)
When I collected the cells after n days of treatment, I carried out PCR targetting a specific gene, then normalised to a housekeeping gene, I carried out analysis using the DDCT method.
I now have the DCT values in a spreadsheet waiting for statistical analysis- however, I am not sure which test to use, or whether I should use DCT or DDCT values.
My DCT results seem to tick the boxes for a non-parametric test, but I am not sure if it should be paired or unpaired. My logic is that they could be paired, since they came from the same cell line, or that they could be unpaired since I am comparing Day n (control) vs Day n (treated).
Results:
Control= 12.59 11.19 12.31 10.89 11.08 11.32 10.97 8.48 10.25
Treated= 11.28 10.98 10.35 11.39 10.36 10.83 10.77 7.91 9.8
DDCT values:
-1.31 -0.21 -1.97 0.49 -0.73 -0.49 -0.2 -0.57 -0.45
Please let me know if you would like more detail. I'm still something of a statistics beginner, so I'm not sure how much depth is required to be able to answer this question.
If you take the data as paired or not should follow from your experimental design. Is there any source of variance that acts on pairs of treated and control samples? For instance, one cell culture was grown, then splitted in two vials, one of which was treated. This gave the first pair of values. Then another, culture was grown independently, split, treated, measured. That would be the second pair etc.
Calculating individual ddCt values makes sense ONLY when you do have a pairing (and when it is given which two values are a pair). Then you would test if the mean of these ddCt values is zero, using a one-sample t-test, which the very same as using a paired-sample t-test on the two groups of dCt values.
If the data are not paired, then you would test the mean difference of the two samples of dCt values using a two-sample (independent-sample) t-test.
It seems to me that choosing two plates from the same cell line brings the design closer to one of paired data. The correct procedure would have been to randomly assign one to treatment and the other to control, for each pair of plates. You know how you did it.
Sounds independent to me.
That said, you could also conduct repeated-measures ANOVA in order to account for your repeated-measures between 2 groups.
Hope this helps,
Matt
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis