Non-zero expression counts from deleted genes

Question

[loughrae]

Hi all,

I've been looking at the RNA-seq data (STAR counts) for homozygously deleted genes (ASCAT copy number = 0) in TCGA and don't understand what I'm seeing.

I expected these genes to have zero or negligible TPMs, but actually the vast majority have TPM > 1. Some are even above the average expression for that gene in two copies.

I've already checked that the full gene is deleted, so RNA-seq shouldn't be picking up expression from a non-deleted part of the gene. Is there a known explanation for this? Is it more likely to be an error in the copy number data or the RNA counts? These are imprinted genes and there are about 60 gene-cases if that's relevant.

Thanks for any help.

[benformatics]

I've already checked that the full gene is deleted

What does this mean? Did you check that the gene is fully deleted in the RNA-seq? What does the coverage look like?

[loughrae]

As in the DNA copy number from ASCAT is 0 across the full length of the gene. For RNA-seq I only have counts.

[benformatics]

Well then you don't know that the gene is fully deleted ;) My guess is that it's partial and that you have residuals counts at the flanks