From Eterna Wiki


Looking for something to do?

Don't you guys love this first sentence you find when opening EteRNA? Well, typically, I never have to ask myself that profoundly philosophical question. But it seems that I won't spend so much time in the general areas of this wiki anymore, so this morning, I was actually looking for something to do...

What about the current labs? Let's see... Single to Switch try #2... doesn't look like I submitted anything yet...

Blog frabase 1.png

Hmmm, I'm not sure that I understand JR's idea and project. But forgetting about that, the problem looks like a fairly simple one, at least in terms of structure. The "challenge" is of course, the multiloop.

I wondered which database would let me look up precisely such a 0-1-4 three-way junction, and I decided to test a database I heard about while playing with the RNAComposer 3D design engine.



The database is located at and it's actually pretty forward to use. 

  • Menu "Structural Elements"
  • Tab "Loop"
  • Sequence lengths 2, 3, 6 (they take the lengths including the closing base pairs, so it needs +2 on every segment length)
  • check "Strand shift operation" (see this help page for what is this option for)

A few hits, nice, mostly identical though... let's study the first one, PDB accession 2JL6. There's a very nifty feature in this output, namely the possibility to visualize the fragment in 3D. Cool.

Blog frabase 2.png

Quite verbose... after fighting a little with the Jmol applet, I got this:

Blog frabase 3.png

Much more comprehensible (for me), but still... Because I couldn't easily color the bases, and also because I noticed some oddities in the placement of some of them, I decided to work the "beast" (I had no idea how much of one it would actually be) locally with Chimera.


Visualizing in 3D

Ok, Chimera fired-up, fetch 2JL6...

Blog frabase 4.png

oh boy... an interesting case of TMI, wouldn't you say? The FRABASE output gave us precise location informations though.

Chain Start End
B 11 16
B 68 69
B 108 110

As I mentioned earlier, some strange arrangement of the loop bases led me to think that something could be going on with the pairs surrounding the loop, so I decided to expand the ranges by one base, in all directions. In Chimera, I did:

  • Menu Select / Atom specifier...
  • :10-17.b,67-70.b,107-111.b

Blog frabase 5.png

Aha, here it is... Now, what about looking at just that segment?

  • Select / Invert (all models)
  • Action / Atoms/Bonds / Hide
  • Action / Ribbon / Hide
  • Select / Invert (all models)
  • Action / Atoms/Bonds / Show
  • Action / Ribbon / Show
  • Select / Clear selection

Next, I colored just like described in the Quickstart I wrote some time ago, and I got:

Blog frabase 6.png

Intuitively, it is feasible to figure out possible H-bond interactions just by rotating and observing carefully the structure, but we can ask the software to find H-Bonds for us.

  • Tools / Structure Analysis / FindHBond

Blog frabase 7.png

And we get:

Blog frabase 8.png


There's an interesting feature about one of the free A's. Try this:

  • Menu Select / Atom specifier...
  • :13.b@O2'

Color it with something contrasting and watch:

Blog frabase 11.png

This is the spot where the SHAPE reagent is supposed to bind with the nucleotide. If the nucleotide is somewhat mobile (not base-paired), the reagent can normally bind there, the nucleotide is declared "reactive", and it appears yellow in our lab results. If not, the nucleotide is declared "protected" and shows in shades of blue. In this specific case though, one can see a possible reason why a nucleotide may appear protected even though it is not base-paired: if the binding spot is enclosed in a net of other molecules, the reagent may have no chances at all to reach it. Which is why, even though this Adenine only binds weakly to a Guanine and doesn't seem to stack much with anything, it will likely appear very protected (quite blue).


Back to the 2D plane

Now, projecting the data we got on the 2D plane gives:

Blog frabase 9.png

And if I add the possible non-canonical interactions that are visible in the 3D model, I get:

Blog frabase 10.png


It is likely that I will submit a design including this pattern. For one, I'd like to know what SHAPE says about it. The prediction would be that, if the loop forms as expected, only A62 should appear very reactive (yellow in the lab view), and all other supposedly free loop bases will have some shade of light blue...



It feels to me like there's something "wrong" in the way we're working at EteRNA. The lab challenges ask us to create "clean" loops. In essence, we are asked to create the most unstable loops, while trying to keep the stacks as stable as possible. To me, it doesn't look like Mother Nature works that way. As evidenced by the documented instance of a multiloop I presented above, loops also try to be as stable as possible.