Best Mirna Target Prediction Algorithms
4
4
Entering edit mode
11.7 years ago
Paul ▴ 760

I'm wondering, for the sake of publication, how it is best to justify the use of one miRNA target prediction algorithm over the other? I'm not sure there's a recognised "best" one?

Is there any index out there that might list them by accuracy or maybe somebody might have compiled a list of which algorithms have the most citations?

mirna target prediction • 6.9k views
ADD COMMENT
0
Entering edit mode

Do you want to predict the target of a miRNA? As to what I know that is the only reliable prediction approach and the de-novo prediction of miRNAs is almost infeasible. Any opinion on this?

ADD REPLY
0
Entering edit mode

I should mention that above I mean miRNA target prediction algorithms, like PicTar and TargetScan.

ADD REPLY
0
Entering edit mode

Hi Paul - I've updated title and tags to reflect your comment. In general, I'd suggest editing your post in a situation like this to help clarify your original query.

ADD REPLY
0
Entering edit mode

Hi, I'm also curious too and interested in thoughts about TargetScan v. microrna.org.

ADD REPLY
3
Entering edit mode
11.7 years ago
Cjt ▴ 370

The answer strongly depends on your input.

  • Are you working with animals or plants?
  • Are you looking for precursor predictions or mature prediction?
  • Are you working solely on sequence level or do you use any NGS datasets or chip techniques?
  • Do you have additional degradome data?

These are just some example questions which need to be considered. Doing so, the large set of tools boils down do a very low number of applicable tools which can be quite easily found by taking a look to currently published biology papers of your field.

ADD COMMENT
0
Entering edit mode

I should have been more specific, I meant miRNA traget prediction algorithms (the data I am using is human). I'm currently working with PicTar and TargetScan, but I'm sort of at a loss as to why they might be better than say Pita or Miranda.

Appologies for the vagueness of my question I'm very new to this area!

ADD REPLY
0
Entering edit mode

Most of my reply/question remains the same: The principles of miRNA-target binding are very different for plant an animals. Thus, you will need a specific tool depending on the species.

Are you trying to do a in silico approach or do you have some additional biological data to validate or falsify your predictions? Some tools already incorporate such information, the easiest example are homologue sequences showing conservation. On the other hand this limits the scope of the predictions...

ADD REPLY
0
Entering edit mode

It's a purely in silico approach, using human mRNA microarray data. I think targetscan for one incorporates evolutionary conservation in it's predictions.

ADD REPLY
2
Entering edit mode
11.2 years ago
Madhan ▴ 240

If the aim of the Project itself is to identify the best target prediction algorithm, then you can take this nature paper as a Model study (http://www.nature.com/nature/journal/v455/n7209/pdf/nature07228.pdf)

Or you want to justify your selection of Target prediction algorithms, then quote this paper as reference and use the three prediction algorithm that they suggested "TargetScan, DianaMicroT and miRanda/mirSVR".

Basically, those algorithms which considered complementarity along with conserved nature and free energy performs well over the methods that considers single characteristic.

ADD COMMENT
0
Entering edit mode

Thanks for the good paper. I would suggest also this one: "A ceRNA Hypothesis: The Rosetta Stone of a Hidden RNA Language?" by Salmena & Pandolfi. I know they rely also on rna22 for target predictions.

ADD REPLY
1
Entering edit mode
11.7 years ago

Please see all of the replies to this question on this forum. Summary: Different algorithms take different approaches and so they are more complementary than redundant. Also, a popularity contest (most citations) is not necessarily an indicator of the best tool to use.

ADD COMMENT
0
Entering edit mode
11.1 years ago
Kris ▴ 40

I have noticed that most of the miR prediction algorithms microrna.org, Targetscan, etc) will predicted the same targets - however it is the confidence scores assigned to the predictions that will differ greatly between algorithms... which is a reflection of how different algorithms might weigh different rules/guidelines of interaction (Free energy, seed pairing, site context, conservation, etc).

A predicted target that scores highly (top 5%) in all algorithms would then be considered high confidence.

ADD COMMENT

Login before adding your answer.

Traffic: 1609 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6