Difference when using difference number of pathways score to compare between 3 groups

Question

[Chris]

Hi Biostars,

I am using GSVA package to get pathway score for a data set with 3 groups. Then I use limma to compare pathway score between 3 groups. I just don't know why when I use all pathways score ( a few thousands), the p value from topTable() of pathway A will be different when I use less number of pathway (a few pathways). I think it should be the same regardless of number of pathways. Would you please have an explanation? Is that because of Empirical Bayes Statistics? Thank you so much!
https://www.bioconductor.org/packages/release/bioc/html/GSVA.html

[Gordon Smyth]

https://support.bioconductor.org/p/9157856/