Hi! I'm super novice to this area and looking to perform a gene level meta-analysis of differential expression using data from GEO as well as TCGA. I'm curious if it makes sense to combine RNA-Seq data from TCGA and microarray data from GEO and if there are any gotchas I'm missing.

This is the process that I'm thinking of using:

1. Label each dataset with entrez IDs

2. Normalize within a study (using affy to normalize Affymetrix data from GEO, limma to normalize Illumina from GEO, and can I just use the preprocessed normalized RSEM values (mRNAseq_RSEM_normalized_log2) from TCGA?)

3. Add contrasts (either disease/normal)

4. For each dataset independently run limma differential expression analysis

5. Combine effect sizes (e.g., using MetaMA, or are there any other suggested tools for meta analysis?)

Does this process make theoretical sense? Anyone have any suggestions?

Thanks for your help!