3
$\begingroup$

My experiment involves adding different compounds to the same cell line and then comparing it to the control to see if there's any difference. I am under the impression that ANOVA is used when the groups are from the same population (for example, people grouped by age) which is not my case? Would it be better to conduct multiple t-tests in this case? Thanks

Edit: I'm adding in one small molecule to cells/a cell line. In total, there are 12 small molecules. I want to see if a specific receptor that are on the cells would bind to the small molecule. The higher binding affinity, the higher the amount of that receptor would be found on the cell surface (what I'm measuring). Done this experiment three times. The control is the 'normal' amount of that specific receptor on the cells. I want to compare the control to see if any small molecule has increased the number of that certain receptor.

Sample data that goes up to Compound 5 (in reality, there's 12 compounds): enter image description here

$\endgroup$
5
  • $\begingroup$ Need some details in order to suggest appropriate statistical analysis: How many specimens in control group? Are specimens with Compound A chosen separately from the same cell line. Similarly for Compound B, etc? Or are Compoinds A, B, etc. added in turn to cells taken from the same specimens in the control group? How many specimens? How many compounds. What do you measure to find 'differences"? Are these measurements normally distributed with mean depending on Compound? $\endgroup$
    – BruceET
    Commented Nov 4, 2020 at 8:52
  • $\begingroup$ Hello Bruce, thank you for the response. No specimens involved. I'm adding in one small molecule to cells/a cell line. In total, there are 12 small molecules. I want to see if a specific receptor that are on the cells would bind to the small molecule. The higher binding affinity, the higher the amount of that receptor would be found on the cell surface (what I'm measuring). Done this experiment three times. The control is the 'normal' amount of that specific receptor on the cells. I want to compare the control to see if any small molecule has increased the number of that certain receptor. $\endgroup$
    – Mark
    Commented Nov 4, 2020 at 9:18
  • $\begingroup$ Three times so there's three replicates. $\endgroup$
    – Mark
    Commented Nov 4, 2020 at 9:45
  • $\begingroup$ How many times do you put each of the small molecules into contact with the cells? (3?) And you do that for each of 12 types of molecule? Are you counting number of bound receptors? If so, would results be a few, several dozen, hundreds? Do you want to know whether the 12 types of molecules tend to have different numbers of bound receptors? Can you give a fake data table that might result from such an experiment? Please edit your Question, Not everyone reads comments. $\endgroup$
    – BruceET
    Commented Nov 4, 2020 at 10:00
  • $\begingroup$ Edited my question. Yes I want to see if the 12 types of molecules have a different number of bound receptors compared to the control. In regards to the results (as depicted in the question), I have already 'standardised' the results so what I did after each experiment (row 1 = experiment 1, row 2 = experiment 2, etc) is: number of receptors in compound sample divided by number of receptors in control sample. So the results shown is the fold change. $\endgroup$
    – Mark
    Commented Nov 4, 2020 at 11:14

1 Answer 1

0
$\begingroup$

From a classical statistics point of view, whether you use ANOVA (here, repeated measures ANOVA) or repeated t-tests is up to you. If you have a priori comparisons to test, you are justified in conducting multiple paired samples t tests with some correction for familywise error. But, it sounds more like you want to start with the omnibus repeated measures ANOVA, then conduct follow-up t-tests to localize effects

$\endgroup$

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.