Quantifying Proportion Of Transcription That Is Mirna Vs Mrna

Question

[Paul]

I'm wondering if I have mRNA microarray data and miRNA data from the same sample, is there any way of estimating how many of say my top ranked miRNAs are being pumped out of the cell nucleus for a given time period versus how many of say my top ranked mRNAs are being pumped out in the same period?

Lets say as an example, in my microarray data, one of my mRNAs has an log2 expression level or 8 and one of my miRNAs also has an expression level of 8, is there for example 10 mRNA molecules for every 1 miRNA molecule, 100 mRNA molecules for 1 miRNA etc?

I'm obviously not looking for anything wildly accurate here just an estimate of sorts.

[Michael]

Microarrays are semi-quantitative at best, so you cannot infer anything about exact numbers of molecules. Especially if you look within-sample and try to compare different genes, any estimate will be very inaccurate due to bias. Across samples, you can compare changes in the same gene.

I didn't get your second paragraph, if both mi & mRNA have same expression value, your best bet is same amount of RNA in both is present, isn't it?

[Michael]

Further, do you use a protocol that is selective for the cellular compartment where the RNA is localized? Your first parag. suggests this. What kind of protocol would that be?

[Paul]

"your best bet is same amount of RNA in both is present, isn't it", wasn't really what I was getting at.

I think the fact that an mRNA and a miRNA have the same expression an a 1->16 scale of two different microarrays doesn't mean that there is the same number of physical molecules of each being expressed in the cell. This question may not be possible to answer.

Answer 1

[Istvan Albert]

As Michael has mentioned in the comments the most prudent approach would be to try to characterize your expression levels qualitatively (increase, decrease, presence, absence) rather than in a quantitative manner (ratios, proportions).

Answer 2

[Larry_Parnell]

I agree with the above (Michael and Istvan) with respect to data from an array. I recall that Vivian Cheung's group has some data on this topic that was presented at ASHG in November in Washington, DC, but I cannot find that in my notes. Perhaps it is something that I sent by Twitter under the #ASHG hashtag... In any event, they used RNAseq to examine other aspects of transcription and I think that method would work much better in Paul's situation where average read depth across the transcript can be used to ascertain relative ratios of two transcripts. In other words, looking at percent of transcription of mRNA-coding vs miRNA-coding vs other(?) genes is not best done with array data.

Note added in edit: A recent paper by Cheung's group in which she demonstrates differential allelic expression with sequencing of transcriptomes (RNA-Seq) in order to identify cis and trans regulators is here.