1
$\begingroup$

Take the following scenario: differential analysis of genes expression between two conditions did not yield DE genes. However, the material for the analysis came from a mixture of cell types. It is expected that some of the cell types are more likely to reflect the difference between the conditions that exists phenotypically. So, an idea is raised to perform de-convolution on the cells, and to perform differential analysis expression for each cell-type separately.

My question is whether this sort of analysis will suffer from multiple comparisons problem (since many cell types are tested).

I will stress that within each comparison, as usual, a multiple-hypothesis correction is applied, since many genes are tested.

Improve this question
$\endgroup$
1
  • 1
    $\begingroup$ Your question is meaningless unless you explain what error rate you are trying to control, for example FDR or familywise error rate. If FDR, then I already answered your question here stats.stackexchange.com/questions/518121 $\endgroup$
    – Gordon Smyth
    Apr 21, 2021 at 12:06

1 Answer 1

Reset to default
0
$\begingroup$

Yes, there is a multiple comparisons problem. If you hypothesis was "within hepatocytes, some genes are significantly differentially expressed", then you could choose a suitable threshold, e.g. p=0.5. If you have 20 different cell types, then that same threshold would probably produce a false positive.

The easy way to correct is Bonferroni. Given that your data is expensive, and you are struggling to get information from it, it is worth coding a Benjamini-Hochberg correction instead.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.