predicted vs. known targets - (Feb/29/2008 )

Hello,

Could anyone please comment on why the number of experimentally known (verified) short RNA - target transcript interactions is so far below the number of predicted interactions? If I'm not mistaken, in C.elegans miRBase lists 18 known targets (http://microrna.sanger.ac.uk/cgi-bin/targets/v5/known_targets.pl) and more than 60k predicted interactions.

Is there a biological, financial or any other reason for this large difference?

1, Is the first number (experimentally verified interactions) limited from above? What is the limiting factor?

2, Or is the second number (predicted interactions) much higher than one would expect? Is it true that the entire precursor sequences of most microRNAs listed in the C. elegans miR-target predictions have been located in the genome of C.elegans? Going one step further, is there any evidence that these miRs are actually transcribed in the given organism?

I have the inconvenient feeling that I am unaware of several trivial facts that everyone else knows...

Many thanks in advance for any helpful suggestions/sources/hints,
Illes

I think that the major problem is the lack of a good high-throughput method of validating binding, since luciferase assays are really the gold standard. So I guess cost and time/effort are the limiting factors. Also, the predicted target list probably includes many false positives, but also misses some true positives too.

In plants most miRNAs are highly complementary to the targets and mediate target cleavage. So targets are identified computationally quite easily and confirmed by a modified RACE that maps the 5' end of the target degradation fragments. There is an exceptional miRNA that represses translation, but its action has been questioned. Yet, this mode of action could be underestimated in plants.
To identify miRNA targets at the protein level a good approach may seem the proteomic analysis which could be performed on miRNA overexpressing tissues or on miRNA knock out mutants. However, at least in plants, this approach is not used. I don't know why. Perhaps it is not sensitive enough. Is it used in the animal systems?

I have experience in 2DE based protoemics approach, and I don't think that this one is sensitive enough to detect changes in miRNA mutants.