Same directional rule for GC-pairs in multiloops

  • 2
  • Article
  • Updated 8 years ago
Here is what I wrote two weeks ago:

I have noticed that there seem to be a rule for the direction of the closing GC pairs in inner loops. In most of the winning designs, the GC pairs turn in the same direction. (EG: GCaaaaGCaaaa... and not GCaaaaCG...)

Now I can explain why different turning CG pairs in multiloops is a bad idea. And why there actually is a wrong direction.

I need you to read the second section written by AnticNoice, which Jee recommended to me: (It's a bit down in the post)

http://getsatisfaction.com/eternagame...

AnticNoise noticed that there were a big energy jump between the stack/quad and multiloop where a design broke.

When Jeehyung looked af Mat's 90% Branches V1 design, he realised that it applied to Mat's puzzle as well and probably as a general pattern. Here is what he wrote in the chatsession with Mat and me, 2-5-2011:

I think I found something interesting in your design Mat.. If you look at the center loop, and 3 stacks that stems out from it, the lower the energy difference between the first energy of the stack and the loop, the better the stack formed. It's quite obvious when you see it in continuous mode..

If you look at Mats design in the blue-yellow mode, then you can see it:

http://eterna.cmu.edu/game.php?myType...

The energy difference between the small loops and their quads (a stack of 4 nucleotides) is fine. Mats design only breaks in the neck – which is where the biggest difference in energy between the multiloop and the neck is.

This difference in energy level between quads and multiloops (if too big) is probably why the multiloops breaks for most of us.

Back to GC-pairs turning in opposite direction to each other. The reason why different turning GC-pairs is a problem, is because that they raises the difference in energy level between multiloop and the stems/arms.

I have made a spreadsheet with values from Mat's 90% design. If I start turning GC-pairs the opposite way, this is what happens to the energy level inside the smaller and the big multiloop:

https://spreadsheets.google.com/sprea...

The more GC-pairs you turn in the wrong direction, the bigger the energy difference between multiloop and stems become – whith means the bigger chance that the RNA-structure breaks. You can play with the GC-pairs in the multiloops in Mat's design and check the energydifferences for yourself. (If you haven't already, you need to turn on the advance enery tool, to see this)

So from now on, always remember to place the red nucleotide to the right, when you put GC-pairs in a multiloop.

(In some RNA designs, like the bulged-star, it will be possible to get away with a few wrong turning GC-pairs, probably depending on the energy level in the nearest quads and the bulges in the arms, as a stabilizing factor.)

You can see my article about same direction GC pairs here, it's a bit down in the document. (search for GC-pairs in inner loops - what I then called what is known as multiloops)

http://getsatisfaction.com/eternagame...
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 498 Reply Likes

Posted 8 years ago

  • 2
Photo of Jeehyung Lee

Jeehyung Lee, Alum

  • 708 Posts
  • 94 Reply Likes
Another incredible observation Eli!

I think the spreadsheet is not open to public. (It's asking you to get permissions)

If this is true, we are facing another design challenge - a lot of repeated patterns add up to chance of misparings. However, seems like the way to minimize energy differences and maximize symmetry is to have lot of repeated patterns.
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 498 Reply Likes
Now the spreadsheet should be public.
Photo of starryjess

starryjess

  • 35 Posts
  • 2 Reply Likes
Hi Eli. I just tried to see your spreadsheet and it still says I need permission.
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 498 Reply Likes
Sorry, Starry, it should be working now.
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 498 Reply Likes
Possible exception to my sameturning GC-pairs in multiloops rule.

When I looked at Berex branches design 6-1, I realised that the neckarea might be an exception from my rule. It might be neccessary to turn the GC-pair closing the multiloop and the neck, opposite to the usual right direction, to make the neckarea stick. As Berex design scored 99% in synthesis, there seem to be a certain allovance to do so. I now remember seing other designs, though not that wellscoring, get away with a twist of the GC-pair here, where neck and multiloop meet.

Berex Branches 6-1
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 498 Reply Likes
RNA fold don't like the opposite turning GC-pair in the neck/multiloop though. Viewable under positional entropy. I would love to know if this design scores 100%, if this pair was swiched?
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 498 Reply Likes
Further proff for samedirection rule for GC-pairs in multiloops

I reviewed the latest branches designs and noticed in Wisdanes design, that a mispairing appeared at each of those spots, where a GC-pair was turned in the wrong directon. Red nucleotide should be to the right in multiloops.

Diminishing returns – 6, 88% synthesis score



Notice the mispairing in each wrongturning GC-pair.

Photo of alan.robot

alan.robot

  • 91 Posts
  • 36 Reply Likes
From the pairing patterns, it seems likely that the GAUG in the left arm is pairing with the CUAC on the right arm, so I'm not 100% certain it proves the wrongturning hypothesis as much as it shows you shouldn't use the same sequence for adjacent arms. . .
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 498 Reply Likes
Yes, you shouldn't use the same sequence for adjacent arms - can I quote you for that in my lab guide? I'm aware that the arms in the top Y have a problem. They violate the rule that it is not always good to have two GC-pairs turn in the same direction too. I just didn't find a design that was clean from other problems. But notice the other places where mispairing occur. They show up together with a wrongturning GC-pair. And I think the wrongturning GC-pairs in the top Y makes the problem with mispairing worse.
Photo of wisdave

wisdave

  • 27 Posts
  • 1 Reply Like
Thanks for the input on my design. I was surprised it performed so poorly, but your critique makes it very clear. I am still a bit confused about the "Right" and "Wrong" way to orient the GC pairs. I have read the hypothesis, but it's not clear to me. Can you explain it again? By saying the guanine should always be to the right, is that in terms of the following the nucleotide numbering sequence? Thanks.
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 498 Reply Likes
Wisdave, no critique of your design. I loved it as it made me see what I was looking for. So thanks for making it. Yes, guanine should always be to the right. What I mean by right and wrong, is that no matter which GC-pair in the multiloop you are looking at (if you view the design from top and tilt :) your head around to look at the pair in the bottom) the red nucleotide in the GC-pair should be to the right and the green to the left. When you turn the GC-pairs like this, red to the right, then you should see the negative energy inside the loop go up. Which is what we want, because it means it will hold the loop together better. Also as I have demonstrated elsewhere, it is not good to have energy gaps. Turning GC-pairs wrong means higher positive energy in the multiloop, creating a bigger gap between the multiloop and arms (which are negative).

And to what I said on wrongturning GC-pairs in the top Y, I did not clearly state that I was refering to another rule I have noticed, about when having two or more GC-pairs in an arm, typically at each end of the arm, it is often better to have them switched in opposite direction.
Photo of wisdave

wisdave

  • 27 Posts
  • 1 Reply Like
Thanks for the info. I understand what you're saying. However, have you looked at the 2 designs that scored 100 in round five? Both of them have guanines turned the "wrong" way.

Maybe Aldo got away with it because each of the 4 arms are balanced with one other. Ding got away with it for the same reason, and he also lowered energy by adding a guaniine to each loop.

They both also have GC's turned the same way in the balanced arms. So, I think my take away is that (in similar shapes), you want to minimize the energy in the loops and if you orient GC's oppositely, you need to have another arm doing the same thing.

In looking at my design in round 1 of the current lab, that is exactly what I have done. I may flip pairs 21/26 and 46/51 and see what that does to the plots for round 2. Thanks.
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 498 Reply Likes
Great observations, Wisdane!

When it comes to Aldos 100% branches design, he have turned a GC-pair in the neck in the wrong direction. Berex did the same with his 99 % design. Thanks for pointing this out, as I was wishing for a 100% design was possible with a opposite turned GC-pair, specifically in the neck.

Where neck and multiloop meet: I think this may turn out as the exception to my rule about sameturning GC-pair in the multiloops. So this rule is only for multiloops connected to arms ending in a loop or multiloop.

When it comes to a "wrongturning" GC-pair between multiloop and neck: If it's good for the formation and strength of the neck, it is allowed, although it's usually safer to avoid.

As to the GC-pairs in the two arms which turn in the same direction in Aldos and Dings design, they may be allowed, as they prevent the arms from being identical. I think the GC-pairs that closes tetraloops, are allowed to mirror each other – this is also a general tendency. And like you say, it is probably best allowed in an balanced design.

You don't have to make all GC-pairs in arms twist, just some. You are first in big trouble if you make them all turn same way.

Here is an example. I know it is not the best, as the arms are also identical and that causes them to pair up. But I think the GC-pairs in the small arms of the upper and bottom Y turning the same way are making things worse. I'll keep an eye out for designs where this problem will be more obvious and isolated.

Here is Starry's branches VI, that scored 89%. First in target mode:

Here it is in natural mode:


I'm curious to know what have the biggest responsibility for the RNA to misfold like this, the identical arms pairing up or the untwisted GC-pairs in the arms. My first guess is the identical arms, but I do think the untwisted GC-pairs do have something to say too.

As to the new assymetric lab design, I looking forward to see what it will do with the ”rule” of twisted GC-pairs in the arms. The loops in the arms, could influence how the rule work. But it sounds reasonable to try a flip the basepairs you suggest and see what it does to the dot plot. Good luck!
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 498 Reply Likes
The tendency with opposite (than usual) turning GC-pairs between multiloop and neck continues in half of the 100% designs - Iroppys "LeTracteur Deux avec asymétrie" and Aldo's "Cyborg 2". I like this consistency for the exception to my rule of sameturning GC pairs in multiloops. The neckarea is a special area. Sameturning GC-pairs is still a good idea, but opposite turning are allowed. So far it seems especially in designs with necks with a low collective energy, (3 out of 4 of the 100% designs, with Ding's mod of Mat's 95% design as an exception. Looking forward to see if this tendency holds.