Hi everyone, I want to use tcga database gene expression data for Differential expression analysis (DEA) in r. there are some studies like ACC(TCGA-ACC) that just contain tumor samples and there is no normal sample for them. In such a situation can we use normal samples of other studies? or is there any way to find differentially expressed genes just with tumor samples of tcga database?

It makes limited sense to combine independent datasets, batch effects will dominate the results. Unfortunately RNA-seq analysis is not done cheap, either proper data are available (or you generate them) or one would need to accept that in this case tumor vs normal cannot be done. TCGA-like datasets are often use to define or compare across subtypes within the cancer entity, not discussing now that a proper definition of "normal" for many cancers is problematic anyway.