1
$\begingroup$

I am reading a paper and in the methods section of the supplementary material, they describe how they compared the two conditions (stimulated and unstimulated) by saying.

Statistical analysis of signaling responses in manually gated bone marrow was performed as follows: For each experimental condition, the arcsinh-transformed median intensity of each of 18 intracellular antibodies was compared to the distribution of the arcsinh-transformed median intensities of unstimulated replicates (n=5) in each of 24 manually gated cell populations (resulting in a total of 432 observations per condition). Independent one-sample t-tests were performed with 4 degrees of freedom and p-values were adjusted for 432 multiple comparisons (Bonferroni). All signaling responses with adjusted p-values below 0.05 are listed in Table S3.

I am not able to understand how they compared stimulated and unstimulated conditions by doing a one-sample t-test?

Both the conditions are performed on the same subject for sure, but they are two different samples (or tubes), so does that mean that they have included the median intensities of the stimulated and unstimulated replicates in the same vector and performed the t-test?

For example in R like t.test(x)

Any clarification would be helpful. Thanks.

$\endgroup$
1
  • 2
    $\begingroup$ Could it be that they paired the samples and did a t-test on the paired differences? Would that be consistent with the rest of the article? $\endgroup$
    – Dave
    Jun 5, 2020 at 17:32

2 Answers 2

1
$\begingroup$

I haven't read the whole paper, but (as a biologist) I would understand it to be what @doubled explains in point 2: for each ab stain and for each cell population, they compare what they get in the stimulated sample vs unstimulated and ask if it is significantly different.

$\endgroup$
1
$\begingroup$

Your general understanding of one sample t-tests is correct. I haven't studied biology since high school, but a few thoughts...

  1. The antibodies and replicates are somehow the same, and the t-test is a paired t-test between each treated and untreated similar 'thing', and a paired t-test is a one sample t-test.

  2. In a lot of biological work I know multiple hypothesis testing is a big thing, and when trying to decide what to even analyze, instead of analyzing a million things, you only keep those that have significant signals. So maybe that one-sample t-test is referring to some part of the process where they decide which signals to keep by just comparing the signal effect to some mean (and thus each is a one sample t-test?)

  3. It's in the supplementary methods, so could be a typo or a mistake. If you understand the biology and that doesn't make sense, and it's not 1 or 2, I would email the authors and ask! There's nothing wrong with doing that!

$\endgroup$

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.