Why there are genes that are not present in TCGA normalized rna-seq data?

Question

[Vasei]

I was doing a comparison between gene expressions for TCGA data obtained from microarrays and rna sequencing. I downloaded normalized data from firebrowse and when comparing gene names in those two platforms, I noticed that there are nearly 1000 genes that are present in microarray data but not in rna-seq data. It's in contradiction with my understanding of rna sequencing because I think that rna sequencing must give us whole transcriptome! So why there are genes that are not present in TCGA normalized rna-seq data?

(Sorry, I am begginner in bioinformatics!)

[WouterDeCoster]

Is the data prefiltered? e.g. all genes with an expression <1 FPKM filtered out? What is the depth to which the data was sequenced? Perhaps the genes you are missing are just very lowly expressed.

Or is the RNA-seq polyA enriched and the microarray isn't, and you're looking at rRNA genes?

Just a few thoughts...

[Vasei]

Thanks. I think that data is not filtered because there are many genes with 0 reads across many samples and also there are exactly 20532 genes in all the cohorts.

I will take a look at preparation methods, but I think both of them are using polyA filtering and measuring mRNA levels.

Answer 1

[Jordan Anaya]

Yes, RNA-SEQ will give you the whole transcriptome, but you then have to take those reads and map them to transcripts. The transcript definitions used by the TCGA when they counted the reads determine how many genes will be in the normalized data from firebrowse. As a result, the transcripts from the normalized data may not be the same transcripts that were present on the microarray chips used.

If you are looking at gene expression you may want to check out my data portal www.oncolnc.org.

[Vasei]

Thank you. That sounds completely correct.