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?
The answer strongly depends on your input.
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.
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!
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...
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.
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.
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.
•
written
by
Larry_Parnell
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.
Log in to answer this question.
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?
I should mention that above I mean miRNA target prediction algorithms, like PicTar and TargetScan.
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.
Hi, I'm also curious too and interested in thoughts about TargetScan v. microrna.org.