p-value and adjusted p-value exactly the same

Question

[mimA]

Hi all,

I'm analysing some microarray data and just calculated p-values using a t-test between 2 conditions (each with 3 replicates; total number of genes 14000). My p-values and adjusted p-values are coming out exactly the same. I've tried both FDR and bonferroni for correction using the p.adjust function in R and the results are the same. Is it possible to have the exact same p-values and adjusted p-values. I'm getting a bit doubtful. Any thoughts.

Thanks!!!

[Ming Tommy Tang]

see my post here on this topic http://crazyhottommy.blogspot.com/2015/03/understanding-p-value-multiple.html

Answer 1

[Devon Ryan]

It's impossible to have the same p-value and adjusted p-value after bonferroni correction with 14000 genes. There must be an error with the code at some point, please post it.

Also, you want to use limma rather than directly doing a T-test.

[mimA]

Hi Devon, Thanks for your answer! You were right there was an error in my code somewhere. I fixed it so now the 2 values are not the same. About your comment on using Limma. I didn't exactly follow. You don't think using just a simple t-test and then adjusting for multiple comparisons is advisable? I used the Affy packages for QC and normalisation.

[russhh]

limma is the GOTO package for microarray analysis. limma borrows information across multiple genes to better estimate the variability for a given gene, it provides a more powerful test that the naive t-test would.

[mimA]

ok, will use Limma, thank you!!

[John]

In general, a test + post-hoc correction is never as good as using a multivariate test to begin with.

[Researcher]

Hi mimA, Can you please share the correct syntax for p value correction, as I am also facing the same issue.

[Devon Ryan]

Please create a new post.