Whole transcriptome analysis nominated primarily miRNA-related pathways. Advice?

Question

[LauferVA]

Hello,

I have been working on a meta-analysis of >1500 RNA-seq samples and >100scRNA samples from 5 studies.

The results generated are both externally consistent (meaning the findings reflect what has been found time and again in the literature) and internally consistent (meaning that other data types from the same patients are consistent with the results generated from the transcriptomic analyses alone). For these reasons, I have confidence in the results in general/overall.

However, there has been one surprise: The pathway and network analyses conducted overall look good. However, there is a massive enrichment of miRNA-related pathways among the most significantly differentially regulated gene sets identified. Indeed, easily half of the top 120 findings relate directly to one miRNA or another, despite that miRNAs make up a tiny fraction of the total number of pathways analyzed...

I have several questions about this. 1) one approach I'd like to take is to study the miRNAs nominated, and determine if they share characteristics of any kind that might help explain why they have been nominated. For instance, I can check the genes in the pathway itself for commonalities, but I'd like to be able to look at many different aspects of their biology. For example:

• suppose I found that all of them contain or target a certain sequence, and that this sequence has been discussed before as a pathophysiologic motivator of the disease state I am studying. This would make me much more confident I am looking at something "real".
• alternatively, suppose they are all known to be expressed in the same tissue, and that that tissue is the one i would expect for my phenotype. Here again, although the evidence is indirect, i would regard the results as much more plausible...

2) Have any of you ever generated similar results? Is this a surprise to you, or not so much? Anything to watch out for in this regard?

3) Finally, and relating back to 1), do you recommend any databases of miRNAs that I can use to organize information about many miRNAs at the same time?

Thank you very much for your time and consideraton. VAL

Answer 1

[Pei]

do you recommend any databases of miRNAs that I can use to organize information about many miRNAs at the same time?

I wish to note that mirbase and targetscan are high reputation database. The former provide some microRNAs annotated as high confidence while the latter provide some evolutionary info of microRNA. Of course, in many cases people are talking about microRNA family.

Also, people may organize info about many miRNAs at the same time into gtf/gff format.

HTH.

[LauferVA]

thank you - checking on these two now!!

VAL

Answer 2

[Pei]

However, there is a massive enrichment of miRNA-related pathways among the most significantly differentially regulated gene sets identified. Indeed, ...

As each miRNA family could target hundreds or even thousands of coding genes, it may not be surprising to find a massive enrichment of miRNA-related pathways among the most significantly differentially regulated gene sets identified.

easily half of the top 120 findings relate directly to one miRNA or another, despite that miRNAs make up a tiny fraction of the total number of pathways analyzed...

This is less clear. I think you must be aware that many individual microRNA could target the same gene cohort because they shared the same seed region. They are from one family. It is less clear how the "fraction" was calculated, which divided by which?

[LauferVA]

Yes I agree with the points you make here.

Suffice it to say that, despite doing many analyses like this one in the past, I have not seen results like this before. I accept what you are saying, but from experience in this area, I have never seen this many miRNA related pathways so near the top of the analytical results nominated. It is anecdotal, I accept that. Thank you for your input :-)