Differential expression level of different exons from RNA-seq indicates what?

Question

[biomedicineman1]

I'm focusing on expression analysis of some synaptic genes by RNA-seq data. The gene has around 50 exons and interestingly it's obvious different exons/regions of the gene have quite different expression level.

Suppose one gene has only one isoform having no alternative splicing, then we're expecting all exons should express equally, right?

If we see different exons have different expression level, my explanations are:

1. This indicates there's another isoform.

   However, for this synaptic gene, I don't see any reads spanning exon-exon junction which can directly tell there's alternative splicing. So this really raises up my second question:

2. What metrics most RNA-seq program look at to call alternative splicing? Coverage difference? Reads spanning exon-exon junction? Or both?

3. If there's only one isoform, is it possible that gene expression can be regulated "locally" by whatever mechanisms, so that only some exons (esp. important protein domains) will be differentially expressed?

Anyway, I think for lots of RNA-seq data, you can see huge difference of exon expression, which puzzles me a lot. And hope anyone could give any thoughts on this.

Thanks a lot!

Answer 1

[Charles Warden]

At least with all the RNA-Seq datasets that I have seen, the coverage is typically very uneven. I think this is often due to technical biases in the library, especially if they are consistent between samples.

Luckily, I think there are some pretty good tools for analyzing differential splicing events, which sounds similar to the idea of a differential expression of an exon (exon skipping, alternative exon use, etc). However, make you have paired-end 100 bp reads, if you want good results.

MATS if my favorite tool for splicing analysis.

MISO is another popular option.