Question

Any Experience With Using Or Building Pathways Containing Micrornas?

7

Entering edit mode

13.8 years ago

Larry_Parnell 16k

I would like to hear of your experiences or suggestions concerning the incorporation of microRNA entities into pathway information. I am not interested in Miranda, Pictair and other such datasets - we have extensive experience with looking at predicted and validated microRNA-mRNA interactions. Now, we would like to incorporate this kind of information into pathways such as those presented at KEGG and Reactome.

Ingenuity has announced recently a microRNA component to their product list. Anyone tried this yet?

An important point to consider, of course, is the microRNA-mRNA interaction is only relevant if the two genes are co-expressed - same tissue, same time. This then leads to tissue-specific and time-specific (developmental stage-specific) pathways. So, things could get kind of complicated.

Another important consideration (added as an edit) is some kind of confidence value in assigning the microRNA-mRNA edges in such a network. Here, one could use the scores from Miranda et al databases, I suppose.

Any insight or ideas will be appreciated.

pathway mirna • 3.5k views

ADD COMMENT • link updated 13.8 years ago by Qdjm 1.9k • written 13.8 years ago by Larry_Parnell 16k

score 5 · Answer 1 · 2011-03-29

We currently have some MSc students working on pathways in WikiPathways. This is to a large extend an exercise to better understand how it should be done.

We expect that it will be a multi-step approach. We indeed use database approaches to decide which miRNA's could be important. These we either add to pathways directly or as pathway extensions in cytoscape. We then evaluate experimental miRNA and mRNA data under specific experimental conditions. This should indeed lead to tissue and disease specific pathways. (And it calls for a mechanism where you use generic pathways with specific views).

The database approach is further also used in automated suggestions to curators. It is still helpful to see what suggestions you could get from the database approaches even if you don't want to add them all.

Technically we are also improving the BridgeDB databases that do the mapping of miRNA data to the pathways.

I think what we do and what we ask are very much in line. Although we still have the same questions you have. Maybe an excellent chance for collaborations.

score 4 · Answer 2 · 2011-03-30

I can confirm that IPA is integrating microRNA data a lot more closely than it has in previous versions, I've just run a bunch of analyses through IPA. The network generation now seems to hang (certainly on the datasets I've just run through) onto miRNAs in the networks as hubs (as opposed to p53 being the normal candidate).

The data was always there in Ingenuity, there just seems to be a shift to displaying/integrating it by default.

Obviously there is some granularity to the filtering of species/tissues used in network generation and perhaps that is now even more fine grained for the miRNA data, but I still don't feel confident relying on these interactions.

From a casual inspection data was hung off experimentally verified interactions rather than the predicted ones (although IPA is currently not letting me log in so I can't confirm this with certainty!).

score 1 · Answer 3 · 2011-03-31

1

Entering edit mode

13.8 years ago

Qdjm 1.9k

Great question!

A couple of other things to keep in mind that aren't widely appreciated:

miRNAs are often expressed as part of poly-cistronic clusters of miRNAs (often with different seed regions). When we've tried to infer function of miRNAs, we found it helpful to group miRNAs into these clusters (or by their genomic location, if cluster information is not available)
Strong target mRNAs sometimes contain multiple binding sites for the same miRNA -- this might help in assigning confidence in interactions.