I am analyzing a few data sets that are on different microarray platforms. If I wanted to use RankProd to find up- and down-regulated genes, I figured I should concatenate all of the expression sets into one giant one, and then run RankProd.
However, since the data sets are on different microarrays, they have different probes and hence I'm bound to have tons and tons of NAs in the expression set. Would it be valid to run RankProd on each different microarray and then select a false discovery rate and retain the significant genes across all studies?
Thank you in advance for any insight :)