Hi all,

I have 2 conditions would like to compare: Alzheimer vs control (sample from peripheral blood using microarray). Two conditions don't have differential express genes. But changes may be subtle at the gene level but more pronounced when looking at coordinated changes across sets of genes within pathways. So I tried gene set variation analysis. Robert (the creator of GSVA tool which I tried on my data) suggested me using fgsea because my data doesn't have differential express genes. However, fgsea requires differential express genes as input. Would you please recommend how I should I process in this case? Thank you so much!

fgsea doesn't require the presence of deferentially expressed genes, it takes as the input the differential expression-based ranking of all of the (expressed) genes. You can just use t-statistic from limma as the input statistic for fgsea.