From Eterna Wiki


Not a knot

I realize that I don't really need to preach here, but seriously, isn't RNA truly fascinating? At EteRNA, we're still only scratching the surface of RNA folding. We are currently focusing mostly on static structures limited to two dimensions, and even this supposedly simple topic is already full of many surprises and subtleties. At a higher level of complexity, we already experimented with riboswitches in the past, and we're told we will be working again on those specially hard to predict beasts in the near future (can't wait!). And there are many more marvels...


Pseudoknots are another favorite of mine, as you probably already noticed ;) Up until a couple of days ago, I had never found a 2D visualizer that would please me, but jViz.Rna just made the cut, and I recommend it.

Not a knot 1.png

I recently posted an article about "energy barriers" on the wiki. Very briefly, it's about the fact that some folding paths require stacks to first break apart before a new one can be formed, and that this process is often energetically unfavorable, sometimes too much so. And I've been wondering for a long time if there wouldn't be a way to make these transitions more easy.


So I designed the sequence you're seeing in the pictures. The top and bottom ones are simple hairpins with close free energies, and the structures are mutually exclusive since they share some nucleobases. If this sequence would be synthesized, the results would be hard to predict. The sequence may reach an equilibrium with a superposition of these two shapes, or one of them could be strongly preferred. It would depend on how difficult it is to jump over the energy barrier. For the sake of completeness, there's a third non-pseudoknotted structure, composed of two short stacks, but I don't expect it to play an important role.

Not a knot 3.png


The twist is that the sequence is designed to be able (in principle) to form pseudoknots, so as to (hopefully) facilitate the transition (as depicted on the right). At this point, I wonder if these pseudoknots would only be temporary, or if they could form a lower energy state and actually stabilize...


The little problem with pseudoknots is that nobody really knows how they work. There are numerous software packages available online that predict pseudoknots, but none of them has achieved a particularly impressive performance. Which is no surprise, since the problem has been proven to be "NP-hard" (in EteRNA vernacular, it's a Brourd/drake178/wawan/hoglahoo/bigbloto/-level puzzle of Hell and Damnation). Just to name a few of them in case you're curious: McGenuspKiss, IPknot, HotKnots and KineFold.


Possibly, the sequence I'm presenting to you here could be the basis of one of those "mutate and map" experiments conducted by tsuname (Pablo Cordero). I wonder if this experimental strategy would be able to tell us which conformations were present in the solution, and in which concentrations. Using that information, it would maybe be possible to fine-tune some parameters for prediction tools, like the so-called "pseudoknot initiation cost" in RNAPkplex (that's the current effort in this domain from the authors of the ViennaRNA package).


If nothing else, I hope these pictures will help some players to better visualize what is a pseudoknot (by the way, these 4 examples on the right are called H-class pseudoknots, the simplest ones), and convince them that there's nothing terribly mysterious about them. So don't feel intimidated or anything :)


-- Nando

(Many thanks to hoglahoo and starryjess for proofreading)

Knotty 1.png
Knotty 2.png
Knotty 3.png
Knotty 4.png
and no longer a knot
Not a knot 2a.png same Not a knot 2b.png