Some techniques involve averaging exon expression levels, but this can be a problem for genes that exhibit alternative splicing, so I'm wondering if there's a more stable region to take to get a more accurate idea for total gene expression. Maybe the 3'UTR?