I wanted to know how to obtain differentially expressed genes in merged dataset of rice(GPL2025)

The conventional answer is that you can't easily combine data from different labs and get reasonable results, particularly if you are trying to compare samples that are confounded with technical variables (lab of origin, cell type, etc). You could read the vignettes for the GeneMeta package, and maybe SCAN.UPC or fRMA (although I doubt there are any frozen RMA data for rice).