Why are transcripts of a same gene expressed differently in same samples?

Question

[grayapply2009]

We recently did RNA-seq for our samples and quantified the expression levels of the 7 transcripts of one gene. We found one of these transcripts are expressed significantly higher (30x) than other transcripts across all of our samples.

Any explanation for this constantly high expression for a single transcript? Why do transcripts of the same gene have different expression levels? Are there any papers about this phenomenon?

Please help me. Thanks.

[grayapply2009]

Another question, we did qPCR to validate the RNAseq results. We found less difference between our "major isoform" and all other isoforms than what we found in RNAseq. Is this just a technical difference or does this have something to do with biology?

[goodez]

Was the RNA-seq sample amplified by PCR before sequencing? You could be seeing 30x increase in the RNA-seq results due to effects from PCR in library prep?

[grayapply2009]

But the ratio should not change, right?

RNAseq: Variant 1 / Variant 2 = 25

qPCR: Variant 1 / Variant 2 = 10

How do you explain this?

[goodez]

Because the exponential nature of PCR, a small original difference in Variant 1 and Variant 2 will be much greater after many cycles of PCR. I believe qPCR will account for this, whereas your RNAseq pipeline may not be accounting for that exponential increase.

Please someone chime in if this is incorrect, but that's how I understand it.

[grayapply2009]

That sounds reasonable, goodez.

[GenoMax]

It is not a good practice to ask new questions in a thread that is already there. It breaks the flow of conversation of an existing thread. You should have posted this as an independent question.

[Devon Ryan]

qPCR generally has poorer sensitivity than RNAseq. Further, you were hopefully checking this on new samples, so you're observing a bit of an expected regression to the mean.

Answer 1

[Devon Ryan]

That's expected and it'd be a HUGE biological problem if cell types couldn't have different major isoforms for a given gene. Google "major isoform" or "isoform switching" for a large number of papers on the subject.

Answer 2

[EagleEye]

There is always possibility of isoform specific regulation in different cell types/context. Recently we also found a long non-coding RNA (publication), where one of the isoform out of 6 plays important role in neuroblastoma.

Answer 3

[Kristoffer Vitting-Seerup]

If you are interested you can systematically analyze your RNASeq data for such isoform switches using my R package IsoformSwitchAnalyzeR which when you have the transcript level quantifications can do the rest of the analysis for - both of individual genes and of genome wide patterns.