Area to discuss Switch puzzle analysis and its relation to lab

  • 3
  • Article
  • Updated 1 year ago
This is a forum area to discuss switch puzzles and their relation to lab.

This discussion sprang out of a discussion between Cynwulf and I on switch puzzles that I shared the other day. I put it up at the OpenTB lab while it didn't really belong there. Omei suggested it getting an area on its own. So in the style of Gerry Smith I hereby name and start the discussion.

Here are what fine insights Cynwulf allowed me to share on his behalf on making of switches.

Switch puzzles relation to switch lab

Cynwulf, switches and repeat sequence


Get your switch game on!
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes
  • happy repetitively switchy

Posted 2 years ago

  • 3
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes

Number of potential states for a switch puzzle


Cynwulfs puzzles and question got me inspired.


He asked the following question:


"I would imagine that there is a proportional correspondence between the number of possible states and the number of shapes you can make with a sequence of length=N. But HOW MANY could you have AT ONCE? I am not sure how to answer that." 


Cynwulf earlier wrote this answer to Astromons question about if real RNA could have 6 states:


Including the standard modified RNA bases and REAL RNA has potentially N^10 states where N is the length of the sequence.


Thus for the puzzle above there could be ~374^10 =53544642343907161646949376 states the RNA has a chance of being in at any time.


For the more standard simplified 4-base RNA model the equation becomes N^4, or 374^4= 19565295376 states the RNA has a chance of being in.


These States are similar to the distribution of electrons about an atom. That is, there are very many potential states, but there are only a few states which the system inhabits most of the time.


Because there are certain states which are more favorable and thus occupied more of the time than not, we speak of states. In the case of electrons these states are realized through the electron orbital models and for RNA secondary structure prediction we refer to Folded States.


In both cases these states do not represent absolutes, but rather model the probabilistic nature of the system being considered.


When we use algorithms to model RNA folding, what the shape folds into is calculated to be the most favorable state and thus the one in which we would expect the RNA to be found most of the time, but the RNA itself is constantly changing between states, only spending nanoseconds at a time in most of them.


The Ligand also does not bind and stay bound (unless specifically designed to do so), but rather forms a bound complex with the RNA that again persists based on the probability of this interaction occurring. While bound the ligand increases the probability that the RNA will be in certain states.


If without the ligand the RNA would be in the State-1 conformation 80% and in the State-2 conformation 15% of the time, then perhaps the ligand would (when bound to the RNA) attenuate this ratio to 55% for State-1 and 35% for State-2.


14 Mar 2017

Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes

Multiloops and internal loops as switch junktions


I have been thinking about if Cynwulfs question may get easier to answer if it gets reformulated or simplified.


jandersonlee mentioned the multiloop as a potential structure in relation to get a structure switch.


"Extrapolating more from the paper might suggest that using a multiloop as part of a switching juncture (hinge) could be useful."


I suspect it may be easier to answer this question, if more limits are put on the question. Eg. to what amount of states are possible when given the following limits:


  • The structure gets repeated

  • The sequence overall gets repeated

  • The sequence and the structure are of the same length


Switch puzzles counts as answers. Hint for Cynwulf. :)



Number of switching stems in multiloops and potential amount of states


I have found myself wondering about the relation between numbers of multiloops and numbers of states, for switch puzzles that have both repetitive sections of structure and repetive sequence.


Here I’m thinking about puzzles made out of pure repeat structures. And solved with repeat sequence for those repeat structures.


Basically what are the relation between stems in internal/multiloops and number of possible states?


Here are some rough guesses. They are by no means final. I would love to see more puzzles like Cynwulfs, looking into things like that.


  • Internal loop - two switching stems - two states
  • Multiloop - three stems - two switching stems - two states

  • Multiloop - four stems - three switching stems - two/three states

  • Multiloop - five stems - four switching stems - three/four states

  • Multiloop - six stems - five switching stems - four/five?


For now there seems to be a limit to how many states a puzzle can have and still use repeat structure and sequence.

Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes

Different state limits with different switch elements?


Now switches can switch over less than a multiloop. They can switch over an internal loop. Similar, just because there are multiloops in switches, don't mean that all their stems are taking part in the switch. Hence it is useful separating the stems counted into switching stems and static stems.


Different amount of switching stems per multiloop may vary when there are many loops. As in the below puzzle, the multiloops closest to the center has the largest amount of stems and switching stems.


Cynwulf puzzle example. The red numbers in the multiloops accounts for the number of switching stems in the multiloops.


Switchiest Switch V



The function of static stems in switches


There are limits to how many stems can be in a multiloop before the multiloop will rather form some other structure. At least for lab. I suspect that simulations will pass more. At least for single state puzzles we have been capable of making sun puzzles with an unrealistic amount of stems in a multiloop. :)


                 

 

Example of multiloop with three stems. Two of them switching. Two states.

                       


Now I wonder if not these 4 ornament sections - that are not involved in the switch - could be replaced by just a longer and static stem. I suspect their only function are to be domain separator between the identical structural and similar sequence domains in the switch and prevent them from pairing up with each other in an unwanted/uncomplete structural manner. (Any manner that destroys the structure repeat.)


I suspect at some point there will be a limit for how many repeat structures one can have next to each other. If they are to be of the same structure and with repeat sequence.



Number of multiloops as state limiting factor


The number of states possible may also depend on the number of multiloops present that can pair up with each other. And how they are connected to each other.


Here are a 5 state puzzle by Cynwulf made of 4 states made with 4 identical multiloops that each have 3 switching stems and 1 state where an internal loop break up one multiloop and the whole puzzle becomes a snake form. I have given each multiloop a yellow dot for highlight.


Eterna and fugue 5 state

                


4 different multiloops with each 3 switching stems (4 identical switch motives) gives 4 state. An internal loop added gives an extra 5 state. (And breaks the multiloop pattern)


I think there may be a limit of how high one can go in number of states while still keeping an identical structure. Now I wonder, if state 1 was made with a multiloop instead of its internal loop, could the puzzle get to 5 multiloops in 5 states?

(Edited)
Photo of cynwulf28

cynwulf28

  • 83 Posts
  • 26 Reply Likes
A lot of information to digest, thank you for starting this thread. I had been talking with wwei23 about switches and states at the same time that we were beginning to talk about all of this and he has created a number of puzzles which address at least one question... for an RNA sequence of length N, how many different states can you create from a single sequence? If you haven't seem his many-state puzzles yet I'd check them out. I think that his puzzles suggest some of what we are likely to find on larger scales. 
Photo of Eli Fisker

Eli Fisker

  • 2224 Posts
  • 486 Reply Likes

Np Cynwulf and thx for inspiring the thread into existence!


I had been out of the puzzle solving business in general for a real long time. Mainly because of lab. :) Just did a couple of puzzle here and there, namely switches.


However I did noticed Wwei23’s multistate puzzles. He has set new standards in Eterna for how multistate a switch can be. :)


I have a few observation on them so far:


  • Just because 9 nt two states can solve together, don’t mean they will both be legal for the 9 nt multistate puzzle with max amount of alternative states.

  • The more states, the smaller structure movement in space?

  • The more states, the more repetitive the base sequence (less base variance)



Why not all 9 nt states are solvable in one big multistate puzzle


For now Wwei23 is testing the limits of how many states there can be in one 9 nt puzzle. So far he has hit 22. (Said for the benefit of the others who may not have found the series yet).


His latest puzzle in the series got me wondering.


I found a structure not among his 22 states: (((...).). (There are several - not all of them could be made stable with the aptamer molecule in the puzzlemaker).  


This one I suspected would not work in the full puzzle. So I made a two state puzzle of it. To show that the structures could be solved.



However this puzzle is not solvable in pure magnet segment style. With pure pyrimidine at the first half of the puzzle and guanine at the last. The sequence that the first state in the above puzzle crave, will not fill the full multistate puzzle of Wwei23. So not all potential 9 nt states are possible for 9 nt switch. At least not all together.


So basically I think there can be made different multistate series over the number 9, not all of them compatible. Not all of them holding the same amount of states. The more multistate, the more they need to be willing to share a repetitive sequence. The more states, the smaller structure movement in space?



Number of domains and amount of states


So far 9 nucleotides are only enough to make a single stem at a time. There are no double domain possible.

I suspect that when we get a multistate puzzle long enough to harbour two stems, that the amount of states possible between the two domains for a moment will jump down in numbers, for the shortest puzzles with a length that allows for double domain.

As the two stems would also put state limits on each other both for structure but also for potential solving sequence. Just as the two strands and the sequence in them, in a single stem 9 nt puzzle do.


Domain size and symmetry folds


Cynwulf has also been thinking about domains and the size of them. But opposite to me, from the larger end of the scale. I bring it here as I like it:



“I made this puzzle with two goals in mind. The first is to show a quirk of symmetries which arise from the sequence itself. In many of my previous Switch puzzles I used sequences which had symmetries of the order 2^n, that is, they had symmetries which were 2-fold, 4-fold, 8-fold, etc. This is why many of those puzzles were ~400 bases long, because that length allowed for ease of use for creating switches of the type just mentioned. I should say here that the symmetries I speak of are part of the sequence, but also can be seen in the way the RNA folds. This emergent symmetry of the structure rather than the sequence is something which I always try to convey in my puzzles.


While puzzles with the desired symmetries are easy to make for symmetries of the order 2^n, other orders of symmetry present challenges. I can certainly create a sequence with the desired number of symmetries, but when the RNA folds, often these symmetries fail to translate to the overall design.


I have two thoughts on why this is. Firstly, the RNA may have a natural proclivity toward lower order symmetries. That is, if I make a sequence with 12-fold symmetry, the puzzle is likely to fold into 2 or 4 structurally-symmetric domains rather than 6 or 12 such domains. Another possibility is that the number of bases I decide to use affects the ultimate symmetry of the puzzle. For instance, if I wanted to make a puzzle with 3-fold symmetry, using 90 bases for the puzzle or even 150 bases would more likely result in 3 distinct structural domains than if I were to use 360 or 480 bases as there are more factors which can divide into these larger numbers and thus allow for the greater likelihood of alternate symmetries being displayed in the structure.”



Playground department


Now it is unlikely to be realistic. But I couldn’t help but muse about a potential prime number system responsible for the amount of possible states per number of nucleotides. Given there are fractal patterns in prime numbers. And RNA is fractal in nature. I suspect in particular so for the switch kind.


Prime image list found in Wikipedia.

Actually I found something even better. An image visualization of prime numbers. It is seriously beautiful. It almost looks like the bow arc view that RNA sequences are sometimes displayed as. It also reminds me of Fibonacci structure.

Infographic: Visualizing Prime Numbers, For People Who Suck At Math

Enjoy! Hope this will inspire something. Puzzles most welcome. Even if I can't solve them, I can still enjoy the switching view. :)
(Edited)
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes

Cynwulf made me capable of this...


Armed with the advice Cynwulf shared on switch puzzles I managed to pull something I would never have thought myself capable of.


Those who know my puzzles, know there aren’t many switches among. I haven’t been the best of friends with the switch maker. My latest attempts are still baby steps, but I’m proud I made it this far. Thx, Cynwulf!


As I hope this will help other like me who struggles with switches, I will try show some of my steps towards getting there along with some images


Inspired by Cynwulfs source of puzzle inspiration, I picked a random viral RNA sequence.


First I looked at full viral RNA genomes. But they were way too long. The shortest sequence I could find was 230 nt. Which was also too long as I wanted something that sparked repeats.


Then I thought, perhaps a gene in a virus. Then it hit me. Viral capsid protein. I wanted the mRNA for that. Because that is a repeat structure when expressed as a protein. And it would be way shorter. I searched the internet and found small sequence that fit the bill of trying to make a switch puzzle with structural repeats.


This is the exact sequence:

GUUUUUAUUUUUAAUUUUCUUUCAAAUACUUCCAUCAUGAGUUCUUCACAAAAGAAAGCUGGUGGGAAAGCUGG


This is the paper I lifted the sequence from:


Nucleotide sequence of 5' terminus of alfalfa mosaic virus RNA 4 leading into Coat protein cistron


The viral capsid protein comes from this specific virus: Alfalfa mosaic virus RNA 4.



How to create a switch puzzle Cynwulf style

1) I picked a real long puzzle. I sorted all the puzzles after length, went to some of the last pages that didn’t have puzzles that were longer than the normal 400 nucleotide limit.


The puzzle picked was a Cynwulf one, Texture E of the gothic letter E.  


I pasted in the sequence 6 times - enough for it to fill the puzzle up. I got this rather pretty structure despite it was not fully repeat structure.


http://www.eternagame.org/sites/default/files/chat_screens/8627_1516303760.png


Here follow the story of how I got to it.


2) Paste sequence into puzzle:


GUUUUUAUUUUUAAUUUUCUUUCAAAUACUUCCAUCAUGAGUUCUUCACAAAAGAAAGCUGGUGGGAAAGCUGGGUUUUUAUUUUUAAUUUUCUUUCAAAUACUUCCAUCAUGAGUUCUUCACAAAAGAAAGCUGGUGGGAAAGCUGGGUUUUUAUUUUUAAUUUUCUUUCAAAUACUUCCAUCAUGAGUUCUUCACAAAAGAAAGCUGGUGGGAAAGCUGGGUUUUUAUUUUUAAUUUUCUUUCAAAUACUUCCAUCAUGAGUUCUUCACAAAAGAAAGCUGGUGGGAAAGCUGGGUUUUUAUUUUUAAUUUUCUUUCAAAUACUUCCAUCAUGAGUUCUUCACAAAAGAAAGCUGGUGGGAAAGCUGGGUUUUUAUUUUUAAUUUUCUUUCAAAUAC


Now it looks like this and have the shape of the original puzzle.


http://www.eternagame.org/sites/default/files/chat_screens/8627_1516307276.png



3) Switch to native state.




4) Beam the puzzle to the puzzlemaker (Right click in the puzzle and choose the Beam option)



Notice the puzzle is now stable in the puzzlemaker.


Actually I didn’t get it stable that easy when I first tried it out. I only figured this step by accident.


What I did earlier was beam the puzzle to the puzzle maker, while it was still unstable. So I beamed the puzzle while it was in the target shape. That meant it turned up unstable in the puzzle maker. Then I had to do these extra steps:


Switch to native state



I deleted the last base directly in the puzzle with the Ø button. (While taking note of which base went missing.)



Then I added in the missing base last in the dot bracket field.



And lastly I filled in that C base at that last base, to make the puzzle stable. Now the former native state is the target state.



Back on track with making a switch puzzle.


5) Now I open a new window with the same starter puzzle. I repeat the process, until I’m left with another stable target structure.


6) I make a weak base somewhere I want a switch. I changed the marked C to a U.




7) The structure goes unstable


Now how I did it first time was to do all the thing with base deletion again. Until I had made the new structure into a target structure. But the easier way to do this is to do the following.



8) Copy the puzzle sequence. Open a new window with the same puzzle. Dump the sequence there. Put the puzzle in native mode. Beam it to the puzzlemaker. And its stable.






9) Now it is time to pull the puzzlemaker. It can be find at top of the page where all our player puzzles also are.  




10) Copy in the structures from the puzzles that holds the structure of your new switch


Structure 1:


...............................((((((.(..(((((.....)))))..)))))))(((((((.(((((((.........................((((((.(..(((((.....)))))..)))))))(((((((.(((((((.........................((((((.(..(((((.....)))))..)))))))(((((..(((((.........))))).)))))........((((((.(..(((((.....)))))..)))))))))))))).))))))).........................((((((.(..(((((.....)))))..)))))))))))))).))))))).......................




Structure 2:


...............................((((((.(..(((((.....)))))..)))))))(((((((.(((((((.........................((((((.(..(((((.....)))))..)))))))(((((((.(((((((......................(((((.((((.((((.(((......))))))).)))))))))...................................((((((.(..(((((.....)))))..)))))))))))))).))))))).........................((((((.(..(((((.....)))))..)))))))))))))).))))))).......................




11) Add in the sequence of the second or first state and solve the puzzle. By adding the molecule where instabilities are and if needed changing bases also.


Sequence

GUUUUUAUUUUUAAUUUUCUUUCAAAUACUUCCAUCAUGAGUUCUUCACAAAAGAAAGCUGGUGGGAAAGCUGGGUUUUUAUUUUUAAUUUUCUUUCAAAUACUUCCAUCAUGAGUUCUUCACAAAAGAAAGCUGGUGGGAAAGCUGGGUUUUUAUUUUUAAUUUUCUUUCAAAUACUUUCAUCAUGAGUUCUUCACAAAAGAAAGCUGGUGGGAAAGCUGGGUUUUUAUUUUUAAUUUUCUUUCAAAUACUUCCAUCAUGAGUUCUUCACAAAAGAAAGCUGGUGGGAAAGCUGGGUUUUUAUUUUUAAUUUUCUUUCAAAUACUUCCAUCAUGAGUUCUUCACAAAAGAAAGCUGGUGGGAAAGCUGGGUUUUUAUUUUUAAUUUUCUUUCAAAUAC





I marked all the bases that were involved in change.




12) Now onto publish the puzzle.


Now I could have made this puzzlemaker tutorial showing just how things went fine all the way. However I decided to show as I actually did. This was what happened the first 4 times I attempted making a switch puzzle this for me new way. But assuming I’m not the only one getting stuck, like this I shall show exactly what happened.



Sigh!



13) New strategy


I took the original viral sequence. Added 1 A’s in front of it and 1 A’s after it. The hope was to get a structure that would not overlap so much,.



UUUUUAUUUUUAAUUUUCUUUCAAAUACUUCCAUCAUGAGUUCUUCACAAAAGAAAGCUGGUGGGAAAGCUGG



AGUUUUUAUUUUUAAUUUUCUUUCAAAUACUUCCAUCAUGAGUUCUUCACAAAAGAAAGCUGGUGGGAAAGCUGGA


I had to try with different amounts of A’s around the original sequence. I made it with 1.


But now I have luck. :)


Here is the proof. I just changed one base to get the unstable structure in second state. And adding in the molecule there allowed me to revert to the original sequence.


Alfalfa mosaic virus RNA 4 - Dedicated to Cynwulf


I had tediously locked up most bases in the earlier puzzle that I thought would pass. I got an overlap message. And I forgot to lock the bases in my new version that did pass.


To make up for this I share how you can solve it. Paste in this sequence. It will give you a sense of the switch. Only thing you need to do is to stabilize base 265.


AGUUUUUAUUUUUAAUUUUCUUUCAAAUACUUCCAUCAUGAGUUCUUCACAAAAGAAAGCUGGUGGGAAAGCUGGAAGUAGAAAAAAAAUAAAGAAAAAAAGAUAAUAAAAAAAAAAUACAAACAAACAAUAAUGCUGGUGGGAAAGCUGGAAGUUUUUAUUUUUAAUUUUCUUUCAAAUACUUCCAUCAUGAGUUCUUCACAAAAGAAAGCUGGUGGGAAAGCUGGAAGUUUUUAUUUUUAAUUUUCUUUCAAAUACUUCCAUUAUGAGUUCUUCACAAAAGAAAGCUGGUUUCUGACUACGCGCUUUUUAUUUUUAAUUUUCUUUCAAAUACUUCCAUCAUGAGUUCUUCACAAAAGAAAGCUGGUGGGAAAGCUGGAAGUUUUUAUUUUUUAUUUUC

(Edited)
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes

Repeat structure and repeat sequence in switches


Cynwulf has made a fine example puzzle.


Switch XLIX: Tying the Sheepshank



Here is his puzzle description:


“In this switch I take advantage of the properties of the sequence used in order to create a switch with a specific feature. In this case, a repetitious motif allows for the switch to be doubled up on itself in a style similar to the knot mentioned. I have locked the core bases which essentially means that this switch is solved for you ahead of time. In watching how this design switches, note which sections move where. Also, keep in mind that there are MANY opportunities for alternative designs due to the number of places the ligand may be placed. Feel free to play around with this switch in puzzle maker, I'd like to see what comes of such an experiment :-D”


I did play around with it and got two different and stable switches - that was untill I got bitten by the structure overlap monster... 



Playing repeat sequence on repeat structure to ease solving


Here is how I solved it:


The way I went about it was putting the puzzle in second state where the core motive is on display. Then I practically repeated it on to every similar structure section and the rest was a question of filling in matching bases in state 1. I think I managed to solve the puzzle keeping all these core sequence repeats identical to the original middle motif in state 2.


Cynwulf confirmed my suspicion and even better said that this could be used for all his earlier puzzles of that type. :)


“All similar puzzles share the elegant simplicity of sequence you have described...though there are likely to be alternative methods of solving these puzzles, there should always be a way to solve using simple repeating patterns.”


But do the bots know?


Now I wonder. Solving repeat structure with repeat sequence is a handy shortcut, that human puzzle makers as Cynwulf has used and human puzzle solvers like me can make use off.


But I strongly suspect that the bots won't see the pattern. That could stand for a test... :)


So my questions are these. Are the bots capable of solving switch puzzles like this one, and if so, are the sequence of the solves as pretty and repetitive as Cynwulf made them, and some of us solved them?


Basically are the bots capable of solving this kind of switch puzzle that:


  1. Is 2 state+

  2. Is a symmetric switch

  3. Consists of repeat structure

  4. Is made with repeat sequence


And do bots solve with repetitive sequence also?



How do the bots behave on switch puzzles?


That question I can’t really answer for now.


Back in the early Eterna days, we could see how 3 different bots fared on our single state puzzles. If they got stalled or not. 


Which allowed us to make puzzles just to test if we could stall the bots. We got pretty good running around corners with them, and tripping them up. :)


It would be nice if we got the bots back on solving puzzles. So we can get an idea what are difficult for them.


How did ViennaUTC do?


To get at least some idea, of bot capabilities with switches, I dug up the list of what puzzles Nando’s bot Vinnie has solved.


I searched the list of cleared puzzles by number of states.


Now I don’t know the names of all of Cynwulfs complex switches. But it looks like the Vinnie has not taken many of these designs down. No surprize there. They are super hard. Probably not just for humans. :)


Here is an overview of what Vinnie has solved of switches up till this moment:


3264 puzzles with 2 states

228 puzzles with 3 states

4 puzzles with 4 states

2 puzzles with 8 states


Which is pretty good.


It seems there is a steep curve when it comes to more states. The more states, the less likely it has a solve.


@Nando, is Vinnie still running? The puzzles I see in its list don't seem to hold any of the most recent puzzles.


Photo of cynwulf28

cynwulf28

  • 83 Posts
  • 26 Reply Likes
I would really like to see how the bots do against simplified versions of the switches that have already been discussed here. Perhaps after Open TB :-D 

<br/>
 As for the drop off for the Bots solving puzzles...I believe this to be true, that their success in solving switches drops off with the number of states, however...we must make sure that we don't forget the relative number of switches of each state for there is also a decrease in the number of puzzles the higher the number of states. I think that in order to confirm a trend we must factor in the population of puzzles of number of states n and then run standard statistical analyses to see if the values obtained are significant. Ultimately to get an accurate view of the situation many more puzzles with n>3 states would need to be made. <br/>
  Currently there are 5177 player-made switch puzzles, of these 4053 (78%) are 2-State, and 1067 (21%) are 3-State. Thus 99% of all puzzles are 2 or 3-State switches. 
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes
These are good points.

The solving percentage around follows the amount of puzzles in a state category for now.

There may be another way to go about this than make many more puzzles. This is something I have been thinking about recently.
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes

Can switch puzzles be split up into puzzle types?



Switches don’t all pull a switch the same way. Just to show an example I picked two of Cynwulf’s puzzles.


Some puzzle types covers the switching central point up inside the puzzle like this puzzle. I have marked the start and the finish base in the puzzle. That starting stem is static.




Where for the below puzzle lay the switching central point bare in state 1. It switches across two of the necks for one of the other states.


Mandala


I wonder if there is a hidden potential fourth state in this puzzle? And perhaps also a 5 and a 6th?



Different kind of lab switches



From lab we know there are different puzzle types, just depending on the inputs. And many of them can be combined


  • Switches with aptamers

  • Switches with RNA inputs (microRNA, biomarker inputs, etc)

  • Switches with MS2


Just to mention a few.



Different kind of single state puzzles


Also just for single state puzzles there are different kind of puzzles. Here are a few of the types.


  • Puzzles with special elements (like loop next to a multiloop, zigzags)

  • Puzzles with repeat elements (2-2 loops)

  • Puzzles with repeat structure (Snowflakes)

  • Puzzles with short stems

  • Puzzles with long stems

  • Puzzles with equal length stems

  • Snowflake puzzles (Symmetry over several axis + sequence repeats)

  • Kyurem puzzles (short stems, similar length, bent and bulges)

  • Spacebar puzzles


The bots were tested at different types on single state puzzles. As for categories like that.

The bots fared bad at most of these puzzle types (Except the more long stemmed ones) where players in general did better with solving.


For background see this blog post.



Why we need to categorize switch puzzles after their type


The bots did not fare equally well on each single state puzzle type. I expect something similar to be the case for switch puzzles. Just it may be some other things that gives problems. We need to figure exactly what.


Simply sorting puzzles in smaller categories of similar type puzzles will make it easier to see exactly what type puzzles the bots tends to get in trouble with.

Not all switch puzzles are going to be equally hard for bots to solve. Hence we need to categorize the switch puzzles after type, to show what type of puzzles bots tend to get stuck on. Are they the same as humans? From what I have seen so far on switch lab puzzles, the answer is no. Just as it was the case for single state puzzles.


Here are some different kind of switch puzzles. There are overlaps between single state and switch puzzle types.


  • Different amount of states

  • Asymmetric switches

  • Mirror switches (Jmf and more)

  • Spacebar puzzles (Will switch bots be just as stuck on these as the bots solving static puzzles?)

  • Zigzag puzzles

  • 2-2 loop puzzles - and other repeat element puzzles (Wawan)

  • Small switches versus large (also look at actual number of switching bases)


Somehow asymmetric and very multistate puzzles seems way harder to solve than multistate puzzles with strong repeat structure and sequence. At least to me.




Switch Puzzles with symmetries


This is just preliminary listing. There are different kinds of symmetries and repeats related with switch puzzles.


Symmetry


  • Translational symmetry

  • Snowflakes (Mirror symmetries) (Ding and more)

  • Fractals (Cody, Cynwulf)

  • Symmetry created by repeat sequence (Cynwulf puzzles)

  • Symmetry created by structure repeats

  • Mandala (Cynwulf)

  • Palindrome sequence (Cynwulf)

  • Different amount of symmetry axis


Repeat causers


  • Prime numbers as repeat causes (Cynwulf)

  • Repeat sequence

  • Repeat structure



These are just some of the puzzle types. What more switch puzzle types are there?


You who have lots of experience with making and solving switches, what switch puzzle categories do you see?




Snowflakes belongs to the switch world



Back in the static days of EteRNA, we got snowflakes stalling the bot algorithms. The bots had much trouble solving static versions of snowflake puzzles.


The bots weren’t as good as human players to vary the solves, to prevent misfolds.


But on switches with repeat structure and repeat sequence, the bots may need to embrace exactly that and reduce sequence variation for an easier solve style. Because repeat sequence and structure are switch causing features. :)


Which reminds me, I once did run NuPack on some static puzzles just to see how it would solve them. It made heavy overuse of GC’s. Totally christmas tree style. I ran it on a big symmetric puzzle of the snowflake kind, that had strong repetitive structure and short stems. But it wasn't a surprize NuPack solved like this. It did the same in labs.


The snowflakes puzzles were the key to identify symmetry as trouble maker for static RNA. Whereas natural RNA switches and our switch lab solves have long seemed to crave symmetry.


Easy memory rule: Symmetry belongs to the switches.


So perhaps these pretty snowflake puzzles really belong more in the switch RNA world than they ever belonged in the static RNA world. That’s if they have long enough stems to not make massive overlaps in our puzzlemaker. ;)


I have revived one of our past single snowflake puzzles and made it into a switch puzzle.


Merryskies Snowflake as switch




Let it snow  - The snowflake challenge



Basically I would really like to see both human and bot going about solving switching snowflakes.


In other words, I wish for a blizzard of switching snowflakes. :)


Let the switching begin...

(Edited)
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes
INFO-RNA - Natural Born Switch Maker


Intro


  • Futile search for a switch solving algorithm yielded interesting results
  • Same old bot - new perspective
  • INFO-RNA is perfect for creating switch puzzles
  • How to get a sequence back from INFO-RNA
  • How to make new target states for switch puzzles
  • Split motifs in groups - benefits switching
  • Question: Will more single state RNA bots be helpful for designing switches?


Futile search for a switch solving algorithm yielded interesting results


“With the current interest in using bio-molecules in nano-technology, the ability to design artificial riboswitches reacting to changes in conditions will become increasingly important. Hence an implementation capable of solving the inverse folding problem for multiple structures is a key development in structure design.”

https://bmcbioinformatics.biomedcentral.com/articles/10.1186/1471-2105-13-260


I wanted to know how bots solved our switches. I was searching for energy models with capabilities with solving switches. So far with little luck. At least when it comes to getting a web server. EternaBot has gained switch capabilities for lab, but I’m not aware of it having an interface when it comes to that.


As I was searching for potential switch folding algorithms, my eyes fell on an old EteRNA acquaintance: INFO-RNA. INFO-RNA was one of the bots that were put at solving our single state puzzles.


Just for the fun of it I decided that I wanted to look at how INFO-RNA went about solving one of the snowflakes.


Those of you who have read my various posts about the energy models behaviour on single state puzzles, know that I have been scratching my head about how the bots went about solving lab puzzles and puzzles too. Doing lots of stuff that will not be legal in nature. Or nature as I know it:  Lab. :)


Things that would create misfolds and mean that the static RNA would not fold into the intended structure. 

However I have just totally changed my mind on one of the energy models: INFO-RNA.



Same old bot - new perspective


INFO-RNA solved the puzzle with no problem at all. No big surprise there.


When I pasted the letter sequence of the snowflake it gave me, into the puzzle again and looked at it, I got super happy.


I saw heavy GC repeats in a supposed to be static RNA design and as something new I loved it. INFO-RNA had designed a perfect switchable sequence.


Misfold is really just another word for switching. ;)

I decided to do exactly what I had been doing the last week or so after Cynwulf got me real curious about switch puzzles. Try make a switch puzzle of it.


The bot solved with heavy GC repeats. Even the GC repeats were placed in a repetitive manner. Things that I have pointed out as problematic if in a static RNA.



INFO-RNA is perfect for creating switch puzzles


However INFO-RNA would be absolutely perfect for creating switching RNA: All I had to do were clean the loops up a little bit and make a couple of GU’s in the stems to make the puzzle go unstable and I had a new target structure (alternative structure - like a second state) for a switch. This were the start of my switching snowflake puzzles. Except the first two, they are all designed by bot. :)


In all cases I could simply run the bot on the ground structure, the sequence it spewed out was so heavy in repeats, that I could easily spark a symmetric misfold.


So to get a switchable structure, all you have to do is give INFO-RNA the structure of a single state design that has lots of symmetry and repeat structures. It goes heavily repeat sequence on such puzzles, making them the perfect starter points for making switches. :)


Sometimes one don’t have to do much more than add in a few instabilities to make a new target state for switch puzzle. Not even touch the loop bases. I just threw in the sequence as is and used the Mutation booster to introduce the instabilities where I wanted them. (I’ll get into detail later of how I did this.) My Saccharomyces Cerevisiae puzzles is one such example.



No promises that these switches will also work in lab. That could be worth a test though. ;)


They don’t harbour my favorite switch repeat sequences: GA and CU rich stretches. A bot could be made to spew these. :)


Basically I would love to see INFO-RNA with a tiny tweak. It should do exactly the same as it is doing on symmetrical repetitive structures, but with the twist that it changes some of the GC’ pairs to AU pairs. And the C and U’s should line up, and so should the G and A’s. And it would be cool if some palindromic tendencies were programmed in too. That would make for max switchability and complementarity between sub structures. 


Plus it should cut down a bit on the G loop boosting as that makes the puzzles too stable. ;) Or practice downboosting. Just a tiny notch. Have enough to make the puzzle stable, but balancing it on the edge of being unstable. Which it already does, but now it is what I like most about INFO-RNA.



How to get a sequence back from INFO-RNA


Here is what I did. I opened INFO-RNA.


I gave it the dot bracket structure for the structure I was interested in seeing a solve on.


(.((((((((...))))..((((((((((((((..((((...))))((((....))))((((...))))..)))))))(((((((..((((...))))((((....))))((((...))))..)))))))(((((((..((((...))))((((....))))((((...))))..))))))))))))))..((((...)))))))).)


I had made a downsized version of Merryskies puzzle, with 4 corners instead of 6.






I set the field Maximal numbers of violations to 0.


I only want a sequence that is an exact solve to the structure I put in, else it may not stabilize in EteRNA. It may not either, as INFO-RNA is neither Vienna nor NuPack - the ingame folding engines. But so far all sequences I have gotten back have made a solve to the structure I had in the first of the Vienna folding engines.



Hit the START button - and wait.


Here is an image of the job output. First I can see what structure the job was run on.



http://rna.informatik.uni-freiburg.de/INFORNA/Result.jsp?toolName=INFORNA&jobID=6956337


My search query (Should expire in 30 days)


Next I get a sequence output that I can lift - In the Direct output for copy and paste field.



Output:


GGACCGGCCGGUACGGCACGCCCCCCCCCCCCCAAUCUCAAAGAGAGGGGGAAACCCCGCGGACACCGCACGGGGGGGUCCCCCCAAUCGGAAACCGAGGGGGAAACCCCGCCCGAAGGGCACGGGGGGAGGCCCCCAACCCCAACGGGGGGGGGAAACCCCGCCGAACCGGCACGGGGGCCGGGGGGCAUCCCCUCAGGGGCGGUGC


(.((((((((...))))..((((((((((((((..((((...))))((((....))))((((...))))..)))))))(((((((..((((...))))((((....))))((((...))))..)))))))(((((((..((((...))))((((....))))((((...))))..))))))))))))))..((((...)))))))).)


(.((((((((...))))..((((((((((((((..((((...))))((((....))))((((...))))..)))))))(((((((..((((...))))((((....))))((((...))))..)))))))(((((((..((((...))))((((....))))((((...))))..))))))))))))))..((((...)))))))).)


Both these structures match up with each other so everything is fine. No violations.


I already had the puzzle in the puzzlemaker. I pasted this sequence directly in the puzzle. The puzzle solved. :)


https://www.eternagame.org/web/puzzle/8568786/

There is some real heavy christmas tree sequence there. Which is bad for something that is supposed to work in single state lab.


You can get INFO-RNA to output alternative sequences solves, but usually it isn’t necessary as the first sequence it outputs generally use so much GC repeats and place them in such a way that the sequence will fit the bill for being a switch.


Sometimes it adds in a few AU’s as well and then it may be an advantage to hand make these into GC’s and make them follow the general sequence repeat pattern of the GC pairs. The more repetitiveness - the merrier - for switches that is.


How to make new target states for switch puzzles


After I saw the switch potential in the INFO-RNA solve, I did exactly what I have been doing to create switches. I tried out making a few GU’s out of GC’s and plus crossed version just to see if they would spark instabilities and new beautiful symmetric structures.


I started to remove some of the dirty loop bases that weren’t really doing anything and could potentially mess up the symmetry for the new target state. Then I added a few U’s to make GU’s here and there. Or removed some hairpin loop boosts. In no time I caught three new potential target states.  


Here is my snowflake catching frog puzzle precursor. Notice the instability in the target icon. Half the puzzle has gone red, which here is a great sign, as it suggest the two branches had made a switch with each other.





Frog target state sequence


GGACCGGCCCAAAGGGCACGCCGCCCCCCGCCCAAGCCCAAAGGGCGGGGAAAACCCCGCGGACACCGCACGGGCGGGUCCGCCCAACCGGAAACCGGGGGGGAAACCCCGCCCGAAGGGCACGGGUGGAGGCGCCCAACCCCAAAGGGGGGGGGAAACCCCGCCGAAACGGCACGGGCGCCGGGUGGUAACCGGAAACCGGCGGUGC


Switching to native state. This is new target state I want to use.



Copy the sequence. Paste it into a puzzle of similar length or paste it to a larger one and delete the extra bases. (as Cynwulf explained to me)


Put the puzzle in native mode. Beam this to the puzzlemaker. And voila, new target shape that is stable.



Then I put both structures in the switch maker for two state puzzles along with one of the 2 sequences and made a puzzle. Which of the sequences you want to use, is probably a matter of taste.


I prefer the one where I introduces the mismatches, as I can then see in my new puzzle, exactly where I put them. I also like to put a highlight marker around the mutated bases in the single state where I made the mutations.


My first snowflake Once upon a snowflake 1 -  took real short time to stabilize. Froggie caught a snowflake took me a couple of days. I also had to change a couple of the triloops to tetraloops.


All 3 target states were derivatives made from destabilizations I had made to the original INFO-RNA solve of the puzzle structure I fed it. The new target structures were very fast and easy to find.



Split motifs in even groups to benefit switching and counter bot limit



INFO-RNA has a structure length limit of max 300 bases.


If one plan a structure with 4 motifs, like eg this puzzle, it can be a good idea to half the structure. So take two of the motifs, run them in INFO-RNA and just double the structure in the puzzlemaker again plus add in the sequence twice. That way you not only beats the structure limit, but you also makes sure that there are real good matches between motif 1,2 and 3,4. Because the two groups are identical, they will come with natural attractions built in. Which is perfect for making a new target structure. That’s how I made this puzzle.




Will more single state RNA bots be helpful to make switches?



Anyway, since INFO-RNA is being so helpful making highly repetitive solves on highly repetitive structures, it is really an excellent tool to create additional symmetric and repetitive solves for puzzles in no time. Not all these target states will be easy stabilizing, but just that INFO-RNA spews highly repetitive sequence is real helpful on its own.


Give INFO-RNA symmetric puzzles with repeat structures and it it will spew switchable sequences. :)


I won’t be surprised either if NuPack should turn out to be helpful on the same account, knowing its tendency to go heavy in GC and to be too high in repeats. Practically all I have complaining about NUPACK and INFO-RNA doing wrong on static RNA, is now perfect for making switches. What a wonderful switch world. :)


Will Vienna and Nupack be helpful on switches too? Help with figuring that out...


Symmetric switches are much easier solving than asymmetric switches. Even should these bots not be helpful making symmetric switches, they can still be helpful. They may be helpful making hard RNA switches. :)


Also are there other RNA inverse folding algorithms that will be helpful on the same account?






(Edited)
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes
How to make new targets states with the Mutation Booster


Creating instabilities


Introduce instabilities into the puzzle you intend to make a switch and see what alternative shapes are sparked. This can be done in several ways.


  • Make a GU of an AU or GC pair in a stem

  • Make two crossed GU’s

  • Downboost the boostspots in hairpin loops, etc.


Like a GG mismatch in a hairpin loop will make things go more unstable, same will a UU. Even just unboosting G boosts will often help. Basically U’s and C’s are your best new friends for creating instabilities. These are just examples, there will be multiple ways of doing this.

 

  • Downboost 1-1 loops

A UU in a 1-1 loop will weaken the area, perhaps to an extend where a switch will happen.


  • Find two opposite base pair spots (Symmetry) in two stems. Mutate them against each other.

I used the origin puzzle to play around in, here is a link to the actual puzzle.Noticed the black marker ring highlights. These are the mutation spots. Where I on purpose made the puzzle unstable to provoke a different folding.


You can do this with the mutation booster or by hand. I used the booster until I figured how to do it by hand. You can also let it run over a whole puzzle to see what instabilities may spring. If the puzzle is unstable beforehand this can work well. Generally I find it fastest to spark larger changes by mutating two times two base pairs against each other.


You can also mutate more bases. As long as you can solve the puzzle yourself afterwards. :)


The above puzzle I did by hand by adding a UU at the one site and a AA at the other site. That will cause instabilities. If you like the shape occurring, copy the sequence and make it a new target. When you have found an alternative structure that you like, beam it to the puzzlemaker.


The background sequence in the above puzzle were just as I had received it from INFO-RNA. Here is that original unmutated sequence.


GCGCCCCCCAAGCCCUAGCGUUUGGGCGUCCCCGUAAAUAGGGGACAUCAACCCCGUCUUCAGGGGUGGGGGGCAGUCGCCCCCCAACCCUAAAAUCAAAGGGGCCCCCGUGUAAAGGGGGCACUACCCCCGCGUAAAGGGGGGGGGGGCAA


Almost translationally symmetry mirrored GC repetition. Cerevisia is born to be a switch



This was the new alternative structure rising from the above instability introduced.




Ways of solving instabilities


I have found the following to work.


  1. Try solve, just as if it was a regular switch puzzle. It is, just with the only difference that you are the maker.

  2. Create a 1-1 loop to deal away with mismatching bases and introduce the molecule to stabilize the newly created loop.


On the later: If you have two mismatching bases in a stem, where you need a base pair and you can’t figure how to solve, then the following will sometimes work. (The stem can be too short for this to work.) Add two single bases beside the mismatching base pair. Delete the mismatching base pair. Fill in the mismatching bases, that will now go into a 1-1 loop where they should be perfectly happy. Things are fine, thats if you are not disturbing anything in the other state/s. :) And should the new 1-1 loop not be stable on its own, and you haven’t used the molecule bonus yet, you know just where to use it. :) That’s how I got my Phoenix rising puzzle stable.




Why making a Ground state of the puzzle you want to make a switch of is helpful


You can save several steps while making switches, just by making and publishing a ground state of the puzzle you intend to make extra states of.


Also it will allow you to run the mutation booster in the puzzle. This will allow you to harvest fine misfolds, that will be potential new target states for a switch. That you can’t do in the puzzle maker.



Worked example with the puzzle Binary Star


I took the structure from my ground state for the binary Star puzzle which is taken from mod of Malcolm's binary star.


I dumped it in INFO-RNA as described in the above post.


Inforna input: (.((((.((((.((((.((((....((((.((((.((((.((((....)))).)))).((((.((((....)))).)))).((((.((((....)))).)))).((((.((((....)))).)))).((((.((((....)))).)))).)))).))))....)))).)))).((((.((((....)))).)))).((((.((((....)))).)))).((((.((((....)))).)))).((((.((((....)))).)))).)))).)))).)



Inforna output: CGCCCCGCCCCAGGGGGGCCAGGAAAGGGGGCCCAGGGGGCCCCGAAAGGGGGCCCCAGG
  GGGCCCCGAAAGGGGGCCCCAGGGGGCCCCGAAAGGGGGCCCCAGGGGGCCCCGAAAGGG
  GGCCCCAGGGGGCCCCGAAAGGGGGCCCCAGGGUGCCCUGAGGUGGUGCCCCAGGGGGCC
  CCGAAAGGGGGCCCCAGGGGGCCCCGAAAGGGGGCCCCAGGGGGCCCCGAAAGGGGGCCC
  CAGGGGGCCCCGAAAGGGGGCCCCAGGGGGGGGGGG


I dumped this sequence in the puzzle.

http://rna.informatik.uni-freiburg.de/INFORNA/Result.jsp?toolName=INFORNA&jobID=5357546


I cleaned the puzzle a bit up, and removed the 1-1 G boosts to help the puzzle switch better.



Then I pulled the mutation booster, marked 4 bases to make sure they would be mutated well against each other. It often takes at least two mutation at once to generate an interesting misfold, aka switch. As with crossed GU’s. Given there are a crossed GC or AU pair there beforehand.



Here is the mutation sequence I started out with:


Starter mutation sequence: CGCCCCACCCCAGGGGACCCCGAAAGGGGACCCCAGGGGACCCCGAAAGGGGACCCCAGGGGACCCCGAAAGGGGACCCCAGGGGACCCCGAAAGGGGACCCCAGGGGACCCCGAAAGGGGACCCCAGGGGACCCCGAAAGGGGACCCCAGGGGACCCCGAAAGGGGACCCCAGGGGACCCCGAAAGGGGACCCCAGGGGACCCCGAAAGGGGACCCCAGGGGACCCCGAAAGGGGACCCCAGGGGACCCCGAAAGGGGACCCCAGGGGAGGGGGG


I put the mutation points at far ends of the puzzle to ensure that I would get a long range switch. Here is one of the mutants I decided to use to create my new target structure.




Notice that the 4 mutations created two mismatches. These mismatches broke the structure and made for a swapping over point. The new shape looked exactly the same.


To watch the new unstable structures created by the mutation booster, swap to native state and go through the mutants. Pick any alternative target state you fancy, to make it a part of a switch puzzle.


New native state



Mutated sequence


CGCCCCACCCCAGGGGACCCCGAAAGGGGACCCCAGGGGACCCCGAAAGGGGACCCCAGGGGACCCCGAAAGGGGACCCCAGGGAACCCCGAAAGGGGAACCCAGGGGACCCCGAAAGGGGACCCCAGGGGACCCCGAAAGGGGACCCCAGGGGACCCCGAAAGGGGACCCCAGGGGACCCCGAAAGGGGACCCCAGGGGACCCCGAAAGGGGACCCCAGGGUACCCCGAAAGGGGAUCCCAGGGGACCCCGAAAGGGGACCCCAGGGGAGGGGGG


I put this in my Snowflake ground state 2 puzzle, switched to native state and beamed it to the puzzle maker. Then I had my new target shape.


Then I pulled the switch maker and put in both dot bracket structures plus my new mutated sequence. I placed the molecules. I then did a few base changes to stabilize the puzzle and published the puzzle as Binary Snowflake 1.


So it seems the more structure and sequence two shapes share, the more likely they are to switch. At least in simulation. But this is similar to lab.


Apparently if you pick two base pairs at mirroring spots at each end of a mirroring puzzle, you get the puzzle opposite. :)


The shapes look the same, but the numbers tell on that a switch has happened.



Puzzle solve:


CGCCCCACCCCAGGGGACCCCGAAAGGGGACCCCAGGGGACCCCGAAAGGGGACCCCAGGGGACCCCGAAAGGGGACCCCAGGGUACCCCGAAAGGGGAGCCCAGGGGACCCCGAAAGGGGACCCCAGGGGACCCCGAAAGGGGACCCCAGGGGACCCCGAAAGGGGACCCCAGGGGACCCCGAAAGGGGACCCCAGGGGACCCCGAAAGGGGACCCCAGGGUACCCCGAAAGGGGAGCCCAGGGGACCCCGAAAGGGGACCCCAGGGGAGGGGGG



Solve




I can do this again with a new set of mismatches a different place and get a new switching snowflake in the same style. As long as the mutated base pairs are opposite each other at the same spot and the puzzle is solved in a symmetric manner, I think this will spark a similar structure.


New mismatches


New target structure


(Edited)
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes
What makes a RNA switch hard?

Practically the opposite of what works well for switch labs. ;) What works well there are repeat sequence and symmetry.




Things that makes for a hard switch


  • Asymmetry

  • Well varied sequence (Lack of repeat sequence)

  • Large shifts in structure between states (lack of shared structures between states)

  • Larger puzzles

  • More states (to a certain degree)

  • Many small and uneven length stems (simulation doesn’t care about them being too long to ever switch in real life)

  • Many varied sizes of elements. (That be internal loops, bulges, smaller loops in particular and multiloops too)




This is a hard RNA switch



What will make a real hard switch puzzle to solve, both for humans and bots (I suspect the later too for now) is asymmetry and greatly varied sequence.


Basically sequence that stems from a natural RNA should be a perfect candidate for a hard to switch puzzle or lab. ;) Probably same goes for structure.


The perfect nightmare switch can be made out of a perfect static RNA design structure and sequence. Its asymmetric, it has varied elements size, varied stem lengths, varied elements and varied sequence all in all. Else it wouldn’t be a good static design.


All it takes to make a hellish switch of such a good static RNA sequence is just mutation points are randomly added to it and new target states are made into a switch. Especially if said static RNA is large. And the structure changes between the states are large. That simulation puzzle will be hard by default.


When structure is varied there are no equal regions for touch up and pairing. (Similar size of elements aids them pair up with each other.)


Example from a most beautiful puzzle Cynwulf created. Notice how the hairpin loop and the internal loop that are present in each of the repeat structures, are used for making two such shapes pair with each other for turnoff in that shape.



Despite it being symmetric, there is nothing easy about Cynwulfs large symmetrical switches.



Discussion, cryptography and trapdoors


I have been discussing hardness of switches with Omei. Here comes part of our discussion. Back then there were still no solvers of my switching snowflakes.


At first I thought that my snowflakes would also be hard to solve for humans. (The single state snowflake puzzles earlier were hard to bots, but not humans) Basically I was new to making them switch, so I didn’t had a feel yet. While not yet in huge numbers, players have caught up with most of them. Except for one of my 3 state snowflakes. They are not easy, but they are not impossible.


However none have yet solved my asymmetric switch monsters that were based on structure from a single state RNA puzzle. (IRES_VEGF_A: As nature never intended it) I didn’t even try with the original natural sequence. I don’t even know if the puzzle structure is 1:1 in size with the native structure. Just that the sequence from the solve itself which were varied a lot (not much repeat bases) seemed to be enough.



Tuesday 23 january


eli [3:01 PM]

Sending you a teaser puzzle made by my new method. It took me very little time doing so whereas I had struggled to make puzzles of those shapes I showed you and intended for the 4 state switch maker. They looked so easy to make, as I had their shape, but were trouble when I attempted to solve them. [Note: I was trying to make the misfold in the sequence by hand - it was easy sparking a new target shape, but harder to make a solve also when I had to pre-fit the whole sequence on both states.]

http://www.eternagame.org/web/puzzle/8567161/


eli [8:01 PM]

As I said, I have found a shortcut to make real sweet switches


eli [8:03 PM]

No one has solved yet. But me


omei [8:06 PM]

Hm.  This may be a good model of a "trapdoor" function, essential to modern cryptography.  It's a function that is easy to compute, but computationally infeasible to invert, i.e. discover the input from the output.


eli [8:07 PM]

Yes, exactly. I was actually surprised how easy it was designing switches and make them stable, when I had first gotten help on cracking using the switch maker.


[8:07 PM] The hard thing is solving them after the fact. [From scratch]

Cynwulf had to greatly simplify a number of his switches for them to even have my interest attempting to solve. Cynwulf still has crazy switch making skills.



Friday january 26


Small intro that is related to the following discussion. 


Cynwulf had put up a puzzle that morning that I think I worked around 1⁄2-1 hour on. I could only get one state stable. The puzzle shape was messy - aka not symmetric.


http://www.eternagame.org/web/puzzle/8570732/


It is long time since I have been so inspired by a puzzle I couldn’t solve.


The puzzle is called Tripping switch, for some reason it made me think of a Tripwire instead.


That said, I am getting more and more thankful for the consideration that Cynwulf shows in many of his complex switches, when he locks up a good part of the switch.


I decided I wanted to make a asymmetrical switch and went back and dug up some of our early challenge puzzles, that were inspired by real and mostly static RNA. After playing around with this and getting some puzzles up I talked with Omei.



eli [1:31 AM] Ok, it is getting bedtime here. But just wish to show you what I managed to make today:



eli [1:34 AM]

I borrowed the bird from Quasispecies, but I made it switch. :)


The original puzzle was called Phoenix rising. And off with me. Wishing you a good day.



omei [2:18 AM]

Whoa!  That's probably the most artistic puzzle I have ever seen!


eli [11:13 AM]

Hehe. I liked the bird, and it had mirrored parts so I created complementarity between the wings with the sequence I put in. (edited)


I have been thinking about what you said about cryptography, trapdoors and snowflake switches. It turned out that people are able to solve them. Cynwulf must have heard my whisper about him being stalled as he took down several after. Malcolm seem to have a specific knack for them. I think it is not the symmetric puzzles that will be the hardest as the more [shared] shapes there are, the more it will lead back to the original sequence. Because of the symmetry puts a lock on how many possible solves there can be.



eli [11:33 AM]

Just like there are only very limited solves to Wwei23 extreme multistate puzzles. Where I think the trapdoor thing can be true is if one take a big asymmetric puzzle, make a mismatch somewhere and stabilize the new target. It will be hard afterwards to see where the change has been made. It will probably be easy to make, but hard to solve.


I haven't tried that yet. But I suspect that it will take far longer solving. That could stand for a test. Me back to find some big natural asymmetric static RNA, and play with it. :)


eli [11:50 AM]

I just did this with a monster of a puzzle. Took me about 5 minutes stabilizing after I had decided on which puzzle and where to put a mismatch. Unfortunately - and fortunately for the rest - this puzzle is way over the 400 base limit. But here is my proof:


eli [11:51 AM]


eli [11:52 AM]

I built it over this puzzle: http://www.eternagame.org/game/puzzle/259159/


eli [12:09 PM]

Because I already had the key - the knowledge of the original sequence for state 1 - plus the knowledge of which base pair[s] I changed in state 1 to get the shape of state 2 - it was fairly easy for me to iron out the small differences that needed to be changed to make the two states both agree with each other. (edited)


eli [5:58 PM]

Basically I'm saying that asymmetric switches will be harder to solve, the bigger and the more states there are.


I have a 3 state monster. Its an asymmetric switch and as ugly as it comes.


http://www.eternagame.org/web/puzzle/8571025/


eli [6:04 PM]

I think the asymmetric switches will be harder, exactly because they don't share any common structure between the states.


Of cause they can be made share parts, as the sequence may form into similar structures. I however tried to avoid that with those I picked for the switch, as I suspect that would be easier to solve. :)


[8:38 PM]

The more asymmetric the structure, but also the more varied the sequence, the harder the switch.


[8:39 PM]

Basically static rna sequences are perfect candidates for hard switches. :)

Structure and sequence wise


omei [8:40 PM]

When you say hard switch, are you including both creation and solving?


eli [8:40 PM]

I mean more solving


[8:41 PM]

Creation is a bit harder when it is asymmetric, but not much. The hell to pay will be on the solver side. Your trap door thing.


omei [8:42 PM]

This does make sense.


eli [8:42 PM]

I still can't say much about creation difference as yet I'm still rather new to them.




RNA switches, encryption and keys


Today I have been watching some Khan videos about RSA encryption (because they mentioned trap door). I think that the mutation points that I introduce in the ground structure (state1) that gives me a new target - are to be compared to prime factorization. Its a key part.




When one is given only the structure then one has no idea where these mutation points were made.


Thats except if it is a symmetric puzzle. Because then there will be repeat structure and repeat sequence. Repeat structure is also a kind of a key. Just a puzzle having 4 similar parts, is a enormous reduction of complexity.


Just having a section of a puzzle fold in a concerted way is already a reduction of complexity. Something I have gone into for lab switches in the post Christmas snakes.


Plus to create symmetric new targets, I typically place the mutation points in a symmetric manner. Just from the new state one will be able to say something about where these mutation points were more likely to be put. You have no chance of that with an asymmetric puzzle.



RNA encryption with Ribozyme Pistol


I am making a puzzle series that I will use to demonstrate this.


I have made 3 puzzles with up to 4 states. They were easy making. Took me around 5 minutes to stabilize each new state.  


First I give the key to the ground state - a repeat of 4 sequences + a specified amount of A’s in between.


For each new state I make, I add 4 changed bases. Sometimes I only change 2, but since they both are in a pair, a pair are affected. So for 2 states there are 4 mutation points, 3 states, there are 8 mutation points, for 4 states, there are 12 mutation points. Each point affecting multiple other bases. Especially because the different states are entangled.


If one takes the sequence from the last puzzle with the most states and use together with the new target, one introduces less spots to stabilize.


The dot bracket structure of RNA can be thought off as a kind of the public part of a key. That in itself put some limits on what sequence is possible, because pairs have to be legal.


I will also put up a ground state of the mentioned puzzle series, as to help the realization that the key sequence alone can solve this. At last in the series I will give the encryption key. Where I lock just my mutation points. Then everything needed to solve the puzzle, is there. Else it will be a hard puzzle despite it has symmetry and is made from repeat sequence.


I will later make easier versions of some of my switch puzzle. For now I am testing my wings. Plus I need real feedback on what is hard or not. I can see that in the numbers of solvers.
(Edited)
Photo of cynwulf28

cynwulf28

  • 83 Posts
  • 26 Reply Likes
In regards to bot testing for switches, I know that ElNAndo888 has created a number of switches which fit the criteria you mentioned Eli, though he may have already tested them. For example, see this puzzle's description: http://www.eternagame.org/web/puzzle/4036433/ 
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes
I asked Cynwulf to further elaborate his point to make sure I understood. He was thinking about asymmetry as I suspected. I here bring his comment back.

"I noticed that many of his designs lack any obvious symmetry. Yes, I was hinting at that in the post, also, I thought that his comment about testing switches with bots (in the puzzle description) was worth pointing out, if only for historical reasons...I wonder how far he got with that project... it seemed to be outside of the more general Labs...though I know that many of the older labs were Independent projects managed by individual players...so I am unsure how it fits in with everything else that was happening on Eterna at the time."

Good point, Cynwulf. Nandos switches do have a lot of asymmetry to them. He sure got Vinnie chewing through a lot of switches.

While I was digging around in the eterna archieves, I hit upon this puzzle maker guide by Nando.

How to create FMN switch puzzles

When thinking about asymmetric switches, as far as I recall, there is a lot of asymmetry in Malcolm's 2 and 3 state puzzles too. They may not be pure asymmetric as I think Nando's group of switches are.

There is another thing I wonder about. I suspect that asymmetric and symmetric switches call for different solve style. Ever since I realized about magnet segments and sequence repeats from the lab, I have regularly tried to solve in a similar style in switches.

The asymmetric switches seems to resist my attempt. Malcolm's in particular. Meaning, they seem to want a more varied sequence solve style. Which is more as what I would expect of a static RNA lab design.

Eg. if I try solve with AU as I typically do first, the puzzle often don't want to tolerate a lot of sameturning AU's, but will rather have them like every other base an A or an U. This typically seem to happen for puzzles where there are big structural changes. Where as puzzles with slides, do need a repeat solve style.

Whereas the off late symmetric switch puzzles to a further degree allow for and need magnet segments and sequence repeats. In other words, they behave more as I would expect a lab switch to behave.

Although Lroppy hinted at that I had taken that game a bit too far on puzzles, when I made a variation of your Three way Switch.

Lroppy sent me this puzzle hint:

"Cynwulf's Three-Way Switch play is actually simpler than what you created in your puzzle Two Way Switch. Like many of your new switching puzzles, I begin by placing one CG in the best energy configuration at the end of a stack closing a multiloop. I then place the Gs up the stack on the same side and do the same for the Cs on their side of the stack. If these placements are on static (non-switching portions) of the puzzle, then I am happy. And if these placements are on a switching portion, I am usually (almost always) presented (in the other states) with stacks that have just one side filled in completely with G or C. I then add then other side, adding Gs to complete Cs and vice versa.
Now - for many of your new puzzles, that is enough to solve them.
However, Cynwulf's Three-Way has a wrinkle. One (and only one) small section does not solve, but instead shows a stack that is wrong because it shows as CC or GG when I need real bonds. I deduced there was an easy way to resolve this stack - alternating AUs. Add some blocks in the bulges and it was solved.

I hope this helps you get to a solve on Three-Way Switch."
(Edited)
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes
What makes a switch harder or easier?

We have worked on switches for 5-6 years now. The forum is loaded with material on what helps dealing with lab switches. As we have shared our solving strategies, our thoughts about what we thought working and discussing why specific things posed problems.


Cynwulf recently pointed out that it is often hard for newcomers to the game to find the relevant material in the forum.


"I'll get to it...Eterna has too much info...too many projects...too much history for anyone to make sense of who hasn't been here from the beginning...and I suspect that it would still prove a challenge even to those veterans."


He got that absolutely right. I have been working on and off for the last year to pull together an overview of some of the switch material of the forum. About what seems to be working for lab switches. Some of this will also work for switch puzzles. But again as with the single state puzzles, the puzzle engines are far less picky about what they allow than lab. (What makes an RNA puzzle hard?) They need to be. ;)


So it will be interesting knowing where the bot engines differs in their approach to switch puzzles compared to lab. Hence it will be helpful testing out which characteristics from lab switches, that the bots seems to be blind to. You can help with investigating that by making different types of switch puzzles, like Snowflakes, Fractals, Symmetric and asymmetric switches etc. For a list of such puzzles see here.


Remember I have pulled together a lot of material, so don’t try read it at once. Spread it out. I have attempted to sum up what I consider some of the most important points in relation to lab switches. Then I link to background material with further explanation. But for all the new stuff on switches, forum is generally the best place to follow and add to the discussion.


This is not and cannot be a complete sum up. Also this materials collection is of cause biased towards what I think works in lab. If you think other things will work for switches - please write it up or pull your own list. :)


I hope this will serve as inspiration for making new switch puzzles and testing what will stall bots and humans on switch puzzles. Plus of cause help with lab.


A few posts will turn up at several posts, as they cover more than one topic.



Lab puzzle puzzle solving strategies



General switch rules sum up



Central Switch lab discussions


Overview of our past main switch discussions. Attempted sorted somewhat in order with newest on top.


Area to submit design for discussion?

Area to discuss Ways to analyze Lab results

OpenTB results

Designing switches for the first round of OpenTB

Blueprint of the MS2 and FMN Riboswitch

FMN/MS2 Riboswitch Structure

Kissing loop riboswitches

Logic Gates

RNA in - RNA out Puzzle Analysis

Cooperativity Puzzle Analysis

The Real Logic Challenge using NuPACK

Our first multi-input puzzle

Johan’s microRNA lab

Switch Scores for EteRNA Switch Puzzles

Round 97 Riboswitch Lab Discussion

Round 98 NG Switches Lab Discussion




WHAT MAKES A SWITCH LAB EASIER?


Here are a list of things that can aid solving switch lab puzzles. I will try go short. A title, a short sum up of the pattern, its relation to static lab and plus a mention of what I think the bot algorithms react to it for now. The latter is based on my switch puzzle solving experience. Last I put links to past forum posts explaining the pattern.



SWITCHES AND SEQUENCE


Base repeats, sequence repeats


Switches: Switches loves base repeats. Lower sequence variation aids switching (Periodic sequence repeats). The more and large sequence repeat regions, the prettier and symmetric the switch. :)


Static RNA: For static RNA repeat bases and repeat sequence on repeat structure was the sure way to not getting a stable structure. Well varied base distribution is less prone to switching.


Simulation:  Switch puzzles seems to care less about repeat bases and repeat sequence than lab. Symmetric switches seems to call for more repeat bases than asymmetric switches.


Past strategy resurfaced

Huge RNA

Periodic repeats in RNA switches - How can they be programmed?



CU and GA rich stretches in switch elements and switches


Switches: Repeat base sequence is not just random repeat base sequence. There is a pattern to the madness. Lab switches and switch elements in particular (MS2, aptamers, RNA inputs etc) seems to be particularly fond of specific kinds of base repeats as CU and GA rich sequences. These repeats spreads to the rest of the switch too.


Static RNA: CU and GA rich stretches in static RNA tended to cause slides and misfolds if not packed away in longer stems. Well varied base distribution is less prone to switching/misfolds.


Simulation: Much less particular about getting CU and GA repeats than lab.


Natural occurring riboswitch holding an adenine molecule. Image from post below


Adventure in Riboswitch Wonderland



Palindrome base repeats


Switches: If sequence repeat is not just done like random GA’s or CU’s, but palindromic I think it can have even more switching ability. Especially when in the aptamer or switching elements. Advantage is that it leaves more orientations for switch element turnoff.


Static RNA: Palindrome were what made hairpins slide or misfold. Sequences like GAAG or CUUC in small 4 base bair hairpins were sometimes enough. In particular if there were more of them.


Simulation: Cynwulf has found simulation to agree. Switch puzzles love palindromes.



Palindromic Switch puzzles

Palindrome sequence




Magnet segments


Switches: Small regions of repeat C’s and G’s gave trouble in static RNA if there were too many, but these are most welcome in switches. Especially in islands of C’s and U’s, plus G’s and A’s.


The strong bases, G and C stands for making the connection and hook up a stretch in a switch. The weaker bases A and U stands for making the region unstable enough to allow the puzzle to switch also.


Static RNA: Such regions in static RNA tended to make the RNA misfold aka switch. :)


Simulation: Cares less about it. Switch puzzles we can often make switch, while solely being solved with AU or GC. Nature I doubt quite so much. However symmetric puzzles seems to crave magnet segments more than asymmetric puzzles.


Symmetric puzzles, asymmetric puzzles and magnet segments

Magnet segments in natural RNA

Reverse engineering magnet segments

Skipping magnet stones on the sequence sea

Part III - Potential switch pattern?

Christmas Snake

Number of RNA inputs and strength of magnet segments

Switching magnet stem

OFF switches wants C magnet landing sites, ON switches wants G magnet landing sites




SWITCHES AND STRUCTURAL CHARACTERISTICS



Symmetry


Switches: Symmetry or identical structure regions has turned up in a good number of natural riboswitches plus in our switch labs too. I took note of the two almost identical single glycine structures in the cooperative glycine riboswitch while we were trying to make our first cooperative switch. They first caused me quite a headache along with the beautiful proteinlike (aka symmetric) structures I saw in natural occurring tandem aptamers. Because of the nature of static RNA I had came to expect asymmetry and associate symmetry with misfolds and bad designs.


Static RNA: Misfolds are really just another word for switch. ;)


Simulation: While symmetry got single state RNA bots on thin ice, one of these bots is now excellent for making symmetric switches. Will switch bots also be on thin ice with symmetry while solving switch puzzles? Help find out by making snowflakes or other of the puzzles listed here.


Image from here.


Protein switches

Huge RNA

Glycine riboswitches

Protein shape versus RNA

Same State blueprint - now with more symmetry

Riboswitches - it’s all about the symmetry...

Symmetric switch elements

Things that sparks symmetry in switches

INFO-RNA: Natural born switch maker



Coaxial stacking


Switches: Omei took note of coaxial stacking in our RNA lab switches being beneficial to help inputs and reporter binding. It has turned out useful in a wealth of different places. Coaxial stacking in itself creates symmetry.


Static RNA: This is one of the features that seems to be beneficial no matter if it is an a static or switch lab design.


Simulation: Doesn’t seem to care about it being there or not. Not needed as in lab.


Omei’s image of stem coaxial stacked with reporter


Is NUPACK's energy estimation of flush stacking leading us astray in the OpenTB puzzles?

Get free royal flush for lab

More on coaxial stacking

On Static Stems = Stability for Switching

Coaxial stacking




Multiloops and internal loops as switch junctures


Switches: jandersonlee mentioned that the JMB paper touches on multiloops as problematic in static RNA and as he said: “Extrapolating more from the paper might suggest that using a multiloop as part of a switching juncture (hinge) could be useful.”


As mentioned in the JMB paper - multiloops as high free energy creatures. Thus they create instabilities near them, something that will help ready stems for switching.


Static RNA: Too many of the same elements of the same size in the same puzzle is trouble. There are an optimal sequence for solving, but repeat sequence and repeat structure provokes misfolds, aka switches.


This may be why we had trouble with mispairing for the Branches lab as this puzzle has very similar length stem and 3 very similar sized multiloops. There were trouble in other static RNA labs having multiloops close too.


Simulation: Element repeats makes the puzzle easier solving for humans. But will the same be the case for robots?


Cynwulf puzzle with repeats structure and sequence reuse. Not easy for humans though.



Switches, unbalanced energy and multiloops

Multiloops for the switch

Making a switch happen by weakening multiloop in unbound shape (in the switching area)

Huge RNA




Long stems are bad switching stems


Switches: Shorter stems are easier to make switching than long stems. The strength of hydrogen bonds are additive. The longer the stem, the less likely it is to split open and take part in an extreme switch. Although weaker bases like GU or mismatches or internal loops inside help the stem split and switch Or as in this cases make the switch glide.



From the section: NUMBER OF ACTUAL SWITCHING BASE PAIRS AND HARDNESS


Static RNA: Too many of the same size stems, especially if short, were trouble and calling for both GC overuse and misfolds.


Penalize long stems in switching area



Bends and bulges as “downboosters”


Switches: Another way to achieve destabilization in areas that needs to get moving, is to introduce bends (bulges) in longer stem regions that needs to get switching. Bends aso functions as protection against too tight bind of the input. Basically bulges can have a similar function to mismatches and GU’s.


Static RNA: Bulges were fine as long as there weren’t too many and they were of different size. Else the best sequence solve pattern for that type bulges started to stray and change to get in to weaker variations.


Image of Whbob’s winning RO design from the below post. He added bulges to long switching stems that were probably too long to be happy to be switching. (red squares)


The structure of a turned off sequence

Things to watch out for



Static stem


If there is extra space in the design that is not needed for the switch, it helps dealing away with the excess bases and put them in a static stem. Static stems as motif spacers. Keeping puzzle parts that should not interfere with each other, from interacting. Sometimes the static stem also has a structural function in relation to the design. See coaxial stacking.


Static stem in the switching area

Static stems and the microRNA labs

Static stems as domain spacers

Static stem in the switching area

Two static stems in the switching area



Equal length stems


Switches: Equal length stems are a gift to switches. That means you have identical structure more places, that can have identical sequence, which makes a crossover between structures both possible and more easy. This is related with structure reuse between states.


Even length or close to even for switching strands is the natural recipe for symmetry.


Static RNA: Equal length stems were poison to static RNA designs. The more equal length stems, the harder to solve with sequence variations. Especially if these stems were short. This is a part of the reason why Freywa’s Kuyrem puzzles killed the single state bot algorithms.  


Switch Simulation: Doesn’t care as long as you can solve. I still somehow doubt that even a switch lab will put up with this much repeat. There are too many potential partner stems.


Cody’s Fractal Flipper


Equal length stems



Switches, repeat elements and structure reuse


Switches: Repeat elements are beneficial in switches. We have had success with double aptamers. Reusing structure between switching states in labs are regularly helpful.


Static RNA: Static RNA labs were more intolerant to repeat of the same size elements as those often would crave similar sequence and would make the design more prone to misfolding. The more element repeats the worse for design stability and bot algorithms.


Simulation: Static RNA bots hates repeat elements. Switch puzzles are easier solving when they have repeat elements, symmetry and repeat sequence. (But do the bots know?)


The puzzlemaker contrary to the lab doesn’t allow double aptamers. Below the image shows the middle aptamer covering for what is supposed to be two aptamer binding sites. The puzzle would under normal circumstances need them.  Except Wwei23 managed to convince the puzzlemaker to hold a molecule in such a way that it covers for two internal loops - both of which would need a molecule to be stable. Way to go!


Image of Wwei23’s switch puzzle Trinary Snowflake


On a curious note

Extra aptamer rescuing broken aptamer

Huge RNA

Minimal change

Reuse of structure between state



Two multiloops, or a multiloop and internal loop close together


Switches: Bringing switching elements close together in space aids them switching with each other.


Multiloops or internal loops are well suited for switches - as they help bring stems or switching elements close together in space. If these stems got repeat sequence and if some of that repeat sequences matches well more than one place, strands will go visit other partners.


Static RNA: Multiloops close together made for pressured designs. These designs made for hard labs. This seems to be true for static RNA labs especially if they also have several similar length stems and the stems being short. Aka it is helping in a switch.


Simulation: Back in the early EteRNA days Freywa pointed out that the bots had trouble solving static RNA puzzles with multiloops close together. In labs it seems to be an advantage to have the switch elements fairly close to each other. But switch puzzles don’t seems to adhere to that rule.


Pressured designs, unconnected designs and normal easy designs

Energetic pressured designs




SWITCHES, BASE PAIRS AND BASES



GU’s, mismatches and crossed GU’s


Switches: GU’s are central to switches. Both when a stem needs to be switching - especially if it is longer. Gu’s and mismatches are both helpful for destabilizing parts lab switch to aid a switch. Crossed GU’s turned out to be particularly helpful for getting long stems moving. Mismatches and GU can prevents too tight binding of inputs. Plus help get longer switching stems moving. Mismatches can function just as GU’s for help get a switching stem moving.


Static RNA: Crossed GU’s worst characteristics is that they split stems open in static lab, which however can be their finest quality in switch labs, where stems needs to get moving.

GU’s however helped in static RNA as they helped introduce sequence variance.


Simulation: In some puzzles a GU or crossed GU’s are also the final part of making a puzzle stabilize. Both for placing a crossed GU n the native state to destabilize it and make the puzzle solve, or place a GU in the target state to stabilize it.


GU-pairs in switch designs

Static versus switch lab - GU’s, mismatches and internal loops

Crossed GU in reporter labs

Crossed GU examples in Logic gates

Use of GU in two input labs

1 base difference - 10% score difference

Breaking up longer stems

A mismatch as switch speeder?

Mismatches between oligo and complement

Number of GU-pairs - Bound/Unbound

Crossed GU’s are now legit :)




Closing base pairs of stems


Switches: Lab switches designs don’t prefer the same base pair distribution as static RNA labs designs. Switches tends gets too unswitchy if all switching stems have closing GC basepairs. So stems in the switching area regularly have have a closing GC pair at one end and perhaps an AU at the other end.


Static RNA: Static RNA generally prefer closing GC pairs for stems most of the itme.

Simulation: Switch puzzles aka bots don’t seem to care to the same degree.


GC-pairs

Closing base pairs



Dirty multiloop rings or overhang bases


Switches: Can aid the switching if played right. Other than A nucleotides at boost spot in the multiloop may however cause stability. But similar to hairpin loops, down boosting may help create slides and complementarity elsewhere.  Where U in static RNA were the worst choice for stabilizing things at boost spots, now it is a good choice for destabilizing. :)


Static RNA: What made the bots often fail in the classic Eterna static RNA designs were their tendency to throw all kinds of bases other than A in the multiloop. The designs need non A bases. But there is a frequency distribution of them. Put in too many C’s and G’s in the multiloop ring and they are bound to want to go on visit elsewhere else they aren’t supposed to. Aka misfold. In switches aka switch. :)


Blue, green and red nucleotides in multiloop ring

Clean halo in multiloop ring or flaws of Vienna




SWITCHES AND ENERGY



Uneven energy distribution


Switches: Switches start show uneven energy distribution. In particular in the labs that have more inputs. It is really the natural consequence of repeat sequence and GA and CU rich sequence in particular. It is probably also helpful to making the switch happen.


Static RNA: Static RNA seems to favor even energy distribution,.


Simulation: Doesn’t care, but probably should. :)


Uneven energy distribution in natural switches

Uneven energy distribution in two input switches

Even energy distribution in RNA designs



Switches and high entropy


Switches: Switches seem to have a higher entropy. Perhaps we can use that somehow as a discriminator.


Static RNA: Higher entropy in supposed to be static RNA was generally a bad sign.


Simulation: Cynwulf regularly shoot very pretty dotplots and repetitive entropy curves from his symmetric switch puzzles. :)

Image found at Cynwulfs puzzle


Repeat bases and entropy

Entropy and switches

Entropy and zippering puzzle parts

MicroRNA and entropy



Downboosting hairpin loops etc


Switches: Negatively boosted hairpin loops and bottom is regularly beneficial to get a switch happen. Similar for 1-1 loops and internal loops. Basically anywhere one has a switching region, may benefit, especially if a switching stem is long or an input bind is long.


Why: The prime function of a regular loop boost in static RNA lab designs is stabilizing the hairpin loop stem. Downbosting a loop will help the loop let go. This is regularly helpful around switching stems. Both at hairpin loop end and at the bottom of the stem. Similarly around the MS2 hairpin. Or around inputs etc. Anything that needs to get moving


Static RNA: Downboosting would introduce instabilities, that could lead to misfolds if the surrounding area in the lab design were not stable.


Simulation: Downboosting in switch puzzle is also regularly helpful.


Image from here.


Hairpin in the switching area

Down boost hairpin loops in switching area

Downboosting

MS2 neighbor bases - continued




SWITCH ELEMENT POSITIONING


Position of switch elements in relation to each other

Switches: If switching elements like an FMN aptamer and a MS2 are too close or too far apart the switch won’t go equally well. ON switches behaves different to OFF switches. ON switches like an aptamer and a MS2 apart, while OFF switches like the aptamer and the MS2 next to each other. For input based labs with reporter input, the ON switches likes the reporter next to the input, while OFF switches like the input distanced to the reporter in the state where the reporter needs to be off.  


Simulation. Switch puzzles aka bot engines don’t seem to care about MS2 and aptamer positions in relation to each other as long as you can solve. (Hint - they should. ;) )


Image from here.


Distance between aptamer and MS2 sequence matters

Distance of MS2 and FMN sequences in exclusion and same state labs

Switch Structure Overview

Get the switching parts close together

Blueprint of the MS2 and FMN Riboswitch

The MS2/FMN riboswitch blueprint still hold up also for other hairpins and aptamers

Blueprint updated




Aptamer has favorite orientation in relation to the switching area


Switches: The aptamer doesn’t seem to switch equally well, just depending on what orientation it has in relation to the switching area of the design.


Image from here.


Different orientation of FMN

Switching parts too far apart or too close

Which aptamer sequence to target for turnoff?

FMN - forward & reverse - orientation matters

Ladybug for FMN Aptamer Orientation Memorization




OFF switches fill more than ON switches


Switches: ON switches don’t tend to need as much space as OFF switches. Two partner labs, on an OFF switch the other an ON switch


ON switch - only uses part of the sequence space


OFF switch - uses more of the sequence space


(Edited)
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes
Ha, it seems I have hit a post length limit :) Here comes the continuation:


Input reversal between ON and OFF partner switches


OFF lab, has its strong input 5’ and weak input 3’


ON switch, has its strong input 3’ and its weak input 5’

Images from here.


Partner labs and input reversal

Input order - depends on if it is an ON switch or an OFF switch

Mirror entanglement still happening




Lane sharing


Switches: But it is regularly useful in switch labs for when two inputs are in competition to each other for landing the same spot in the sequence. That way they can take turns pushing each other out. Depending on concentration.


Simulation: This we can’t do in the switch puzzles as they don’t have RNA inputs.


Lane sharing in EternaBot solves

One landing lane, 2 planes

[A]/[B] discussion - Lane sharing



Salish’s End bits


Endbases affect switch lab design - due to them being next to the DNA scaffold holding the RNA design. Stems put close to the end of the designs can get involved in coaxial stacking.


Salish’s end bit discovery put in perspective

End bit graphs

End bits continued



Input order matters

I’m still not sure if this is an artefact of our fluorescent tag or if it is a phenomenon on its own. But since I have seen on bigger scale that input order matters, for if a switch is an ON or OFF lab, I still suspect it may. When one make inputs overlap, their order of overlapping seems to matter.


Grammar of the RNA

RNA as a Teeter tooter

More on input order

Prediction for round 1 simple RIRO tb labs

Result from prediction on input order

Put leaving input before in order than staying input

Leaving sequence before staying sequence



Design complex position matter


I can’t say what is the final best placement, just that it seems to matter for switches that don’t fill the whole sequence space, which end of the sequence their active switching region is put at.


The position of the design complex

R3 experiments



SWITCH ELEMENT MOVING


Impossible switch element position


Switches: Switches are harder to solve if they have their switching elements too far apart or too close - depending on they are an OFF or ON switch.


Simulation: I don’t think the bots notice. But they should.


Is the position of aptamer too far or too close to MS2? I think it will be good to have groups of puzzles sorted after the distance between MS2 and the aptamer and have them run both in simulation and lab. I think two won’t agree on what is optimal. I strongly suspect that humans and the bots will be able to solve switch designs that can never be made to work in lab, due to their switch elements being positioned at impossible distances to each other.


Switching parts too far apart or too close



Conserved sequences


Patterns have shown up in switch labs, where there are a strong conserved sequence around a switching element, like an aptamer, for turnoff of that aptamer and an eventually other switch element like MS2. They sometimes share turnoff sequence. They are also particular about their placement in relation to each other. In other words, switch elements seems to be particular about their surroundings.


Image from below post


Advantage of partial moving switches

The aptamer matrix



Aptamer moving patterns


Switches: Aptamers in switch labs tends to like to have one end of the aptamer closed of with a static stem.


Static RNA: Craved solves that didn’t included magnet segments nearby the aptamer that matched the magnet segments in the aptamer. CC’s near an FMN aptamer with two Twin G’s inside it, caused misfolds.


Simulation: Switch puzzles can be solved in all kinds of manners in relation to this. Plus with illegally long aptamer gates. I doubt they take this into account.


One static aptamer end

Static aptamer end

Which FMN sequence to catch?

When to make the tail pair



Half-(moving)aptamer


Switches: Some aptamers do not have a totally static end at one end. They have a small bit of their surroundings moving as well. They seem to have the advantage of being able to make a bigger switch, but in a good way, since they are still tied up at one end, just further apart, so it is still fairly easy to make the aptamer reform again, when the molecule is around.


Image by Omei from the below post



A promising structure-based design pattern



MS2 turnoff


MS2 is quite strong. It needs real good convincing to get turned off.


Simulation: Is too easily fooled about MS2 turnoff


Image by Omei from the post right below.


Lessons from the past - designing switches with the MS2 hairpin

Shared turnoff sequence

Bachelor’s dilemma

MS2 Turnoff patterns

Trap that MS2

Input and reporter turnoff

Position of the MS2 turnoff sequence in microRNA

AND logic gate - MS2 Turnoff Sequence Recycling

Using MS2 as MS2 turnoff



Base overhangs - Dangle bases


Switches: MicroRNA labs loves dangeling tails of complementary bases to inputs laid out as a trap.

Image from Salishs mod of jandersonlee’s winning design.



MicroRNA welcoming dangling tail

JL’s microRNA fish hook 1

A landing spot for the microRNA

MicroRNA seed region

Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes

Fractal growth of switch puzzles prohibited in the puzzle maker


There is something that I have been thinking about lately.


I was starting out making switch puzzles of two different structural parts. One in each state. I doubled the puzzles by swapping over the structure from each state to the other.


I had expected to be able to continue making bigger and bigger shapes. However I hit a wall. A very specific wall. Omei pointed it out to me before I got there. Here is his point:


The puzzlemaker is prohibiting the puzzles to have fractal growth.


Let me illustrate.


I was talking with Omei and Wwei23 about it earlier over a period of time after I had made my multistate puzzle.



Jan 27 2018


omei [8:29 PM]

It seems possible that there are other ways, more complex than simply repeating a sequence, to create even more states.


eli [8:30 PM]

I really like that thought.

I wonder about fractal growth

I have made a puzzle, that seems like budding yeast.


omei [8:32 PM]

Think of dividing the sequence into two halves, and using a different repetitive sequence in each half.  Repeat this strategy on each of the halves, keep going, and yes you have a fractal.


eli [8:32 PM]

Yay!

I had been thinking of such thing. I did it with structures instead

So you confirm my suspicion

I started with this puzzle


eli [8:34 PM]



eli [8:56 PM]

That puzzle I posted above, I took half from each state and added it to the other and got this:




eli [8:57 PM]

First I tried double just each state in the first puzzle, but I couldn't

I couldn't either add in the extra parts just where I wanted. It seemed they demanded that I put them next to something that was different to themselves

I continued one time more and ended with this:


eli [8:58 PM]


eli [8:59 PM]

And I suspect I may be able to do it again.

So it seems to be important to mix two different structures with each other.


omei [8:59 PM]

I'm not clear on what you can't do.


eli [9:01 PM]

I tried repeat the same structural pattern


eli [9:01 PM]


eli [9:01 PM]

For each state, but it didn't work

However when I did a crossover of the two states as addition, I could make a switch

So each state got a molecule

While there isn't much sequence difference in the solves, they are different over the middle of the puzzle.

So I think one can make fractals two different ways. Have two different elements or subsections of puzzles next to each other, repeat or have two different sequences next to each other, repeat,


omei [9:07 PM]

I'm not sure whether this is relevant, but my proposal for increasing the number of low-energy foldings exponentially wouldn't work for our in-silico puzzles.


eli [9:07 PM]

Why?


omei [9:08 PM]

In order to "show" all the states as separate, we would need to ability to place multiple aptamers at the same time.


eli [9:09 PM]

I am not sure I understand

I mean I could use some extra molecules in the same state. ;)


omei [9:12 PM]

Take the simple case, dividing the RNA into two halves.  Create a two-state puzzle in each half.  You now have 4 states.  But you can only demonstrate 3 of the states in our puzzles, because you would need 2 aptamer instances at the same time to put both halves into their respective bound states.


eli [9:12 PM]

Oh

I took one puzzle I had made (not published yet, where two similar motives were switching with each other. I doubled them by the same approach as my puzzle above. Here is the result:

http://www.eternagame.org/game/puzzle/8571676/

One aptamer for each.

But I think you have pointed out why I can't double it more

The puzzle ended sliding a bit between states, which were not intended.


omei [9:15 PM]

This is a very real limitation for the OpenTB puzzles -- we're creating RNA sequences that have more binding sites for the oligos than the puzzle supports.


eli [9:15 PM]

I made a mistake


Found a new way to illustrate switches


eli [9:17 PM]


omei [9:18 PM]

Nice!


eli [9:19 PM]

By adding the two puzzles together in two states I avoided extra molecules.

"we're creating RNA sequences that have more binding sites for the oligos than the puzzle supports" I think I get it


omei [9:22 PM]

I've been thinking about a new puzzle UI that uses a simplified energy model but would actually show all the potential binding sites for the oligos.


eli [9:23 PM]

Interesting

as is we can have hidden binding sites and not even realize


omei [9:24 PM]

The idea would be to make use of that to do a fast evaluation of each change, balance the energy in that simplified model, and then submit it to NUPACK.

NUPACK actually supports multiple bindings; it's a limitation of the puzzle UI to not show them.


eli [9:25 PM]

Interesting


omei [9:26 PM]

And of course, NUPACK doesn't recognize the energy boost where oligos form adjacent stacks.  So it would only address part of the current weaknesses.


eli [9:26 PM]

I mean I have designed using two reporters - but only one of them binds


omei [9:26 PM]

Right.


eli [9:26 PM]

It would be most useful seeing what actually binds

Also when one is experimenting and want to show what one is doing


omei [9:29 PM]

But you have also designed with multiple input binding sites.  Not that I have one particularly in mind.  But by comparing the the results for the experimental conditions that involve only one and two inputs, you can see where one input binding can create a new binding site for another oligo because of the boost provided by adjacent bindings, i.e. coaxial stacking.


eli [9:30 PM]



eli [9:31 PM]

This would never show as is, even had I got state 4 stable.


omei [9:32 PM]

Right.



January 28 2018


eli [3:15 PM]


I could sure use an extra molecule. I have 0.2 kcal difference between the misfold and the target state and an extra molecule would so much solve it:


eli [3:19 PM]

I wonder if it is solvable without that extra molecule?


eli [3:24 PM]

It is exactly the state that has two landing spots for molecules where I can get the closest in kcal between the misfold and the target shape.


eli [4:24 PM]

I have run into a similar problem when doubling the motif in my puzzle Switching regions 5. I can stabilize either of the states but not both at once.


eli [4:24 PM]


eli [5:35 PM]

I should probably try make more states to get around this


eli [5:44 PM]

Ha, I think wwei23 somehow managed to hack the puzzlemaker to do this extra distribution of the molecule. He put it in a multiloop instead. When I beam my solution to the puzzlemaker and try make it stable, it is impossible. At least with my usual methods. (edited)

Here is the puzzle in the puzzlemaker. I put the molecule in the internal loops as usual. No matter what, I'm always one molecule short. Just as my above puzzles.


eli [5:50 PM]

Here is a link to the puzzle

Link: http://www.eternagame.org/game/puzzle/8574442/


[Added a different image as to not show the solve to wwei’s solve. Notice the molecule inside the middle multiloop. Normally it should have been placed in the two internal loops and two molecules would have been needed.




eli [5:57 PM]

I was starting to think very much about a certain unfortunate prevotella puzzle. That was energetically impossible to solve. (edited)


eli [6:17 PM]

I have managed to put a molecule in at a bulge, put a base lock on it. Then remove the basepairs infront of it. Put in back the basepairs behind it and get the molecule belong to the multiloop. :) I have no idea if it will actually work in a puzzle. But I can see Wwei can do it. Thats inspiring. :)


eli [6:26 PM]

And yup, he hacked the puzzlemaker and he has just spilled how he did it.

:)

http://www.eternagame.org/web/puzzle/8574442/

We need an easier way to do it.... :)




29 January


Here is what I commented at Wwei’s puzzle Trinary Snowflake


Eli Fisker   It is the same reason why I can't add an extra motif to my puzzle Saccharomyces Cerevisiae: As nature never intended it. 5


Here is my best attempt so far:



State 1: ..(((((((..((((........))))((((((.......))))))....(((((.......))))))))))))....(((((((..((((........))))((((((.......))))))....(((((.......))))))))))))....(((((((...((((....((((((.((((((.......))))))....(((((.......)))))(((((((....)))))))...)))))).....))))((((((.......))))))....(((((.......))))))))))))....(((((((..((((........))))((((((.......))))))....(((((.......))))))))))))....(((((((..((((........))))((((((.......))))))....(((((.......))))))))))))..


Structure 2: ..(((((((...((((....((((((.((((((.......))))))....(((((.......)))))(((((((....)))))))...)))))).....))))((((((.......))))))....(((((.......))))))))))))....(((((((..((((........))))((((((.......))))))....(((((.......))))))))))))....(((((((..((((........))))((((((.......))))))....(((((.......))))))))))))....(((((((...((((....((((((.((((((.......))))))....(((((.......)))))(((((((....)))))))...)))))).....))))((((((.......))))))....(((((.......))))))))))))..


My sequence that has a 0.2 kcal difference between target state and misfold in the second state. I think state 2 needs a second molecule, which we don't have:

AAGCCUGCCAAGCCCUGAAUUUUGGGCGUCCCCGAAAAAAGGGGACAAAAACCCCGUAAAAAGGGGUGGCAGGCAAAAGCCUGCCAACCCUAAAUAAAAAGGGGCCCCCGAAAAAAGGGGGCAAAACCUCCGAAAAAAGGAGGGGCGGGCAAAAGUCUUCGAAUCCCUAAAAUUUGGGAGUCUCCGAAAAAAGGGGACAAAAUCCCCGAAAAAAGGGGACGGGGGCAAAAGUCCCCGAACCUCGAAAAAAAGGGGGCCCUCGAAAAAAGGGGGCAAAACCCCCGAAAAAAGGGGGCGGGGAUAAAAGUCUCCCAAACUCUAAAGUUUGAGUGUCCCCAAAAAAAGGGGACUAAAACCCCGUAAAAAGGGGUGGGAGACAAAAGUCUCCCAACUUCGAAAAAAAGGAGGCCUCCGAAAAAAGGAGGCAAAACCACCGAAAAAAGGUGGGGGAGACAA


I doubt it is possible solving. I suspect it is akin to the past prevotella puzzle that were energetically impossible. This puzzle doesn't have a central multiloop between the individual shapes, so I don't think Wwei23's trick can be pulled this time. I will love to be proven wrong and see a solve to this puzzle. Whichever way.

(Edited)
Photo of Brourd

Brourd

  • 452 Posts
  • 82 Reply Likes
I believe the tricks that result in manipulating the loops that a 'molecule' free energy bonus is applied to, should solve the above puzzle, given it changes the way the entire MFE structure is handled (from a puzzle's point of view).

I think one possible solution to the dilemma of expanding the number of switching motifs in the puzzle is to increase the total number of puzzle states that exist. Currently, your puzzle has three independent domains that all switch between two different structures. Additional states would allow for the stabilization of these additional domains. Granted, the addition of multiple 'molecule' bonuses to each state would serve the same purpose, so there is always room for improvement of the game.
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes
MicroRNA based switches


Discussion between Omei and I from today. Added a few puzzle links as I made the puzzles later. 

Tonight I have been playing around starting from the question of what a switch would look like if based on a microRNA.
I took our old friend hsa-mir-208a and it sparked a pretty structure. I think it is perfect for creating a massive 15 state multistate puzzle. :) [Correction 16 states: https://www.eternagame.org/web/puzzle/8599451/ )

omei [12:05 AM]
I was thinking we used a short sequence from hsa-mir-208a as an input oligo. Have I forgotten something?


eli [12:06 AM]
That is correct

omei [12:07 AM]
For the current fun, are you using just that, or the whole mir?

eli [12:07 AM]

https://www.eternagame.org/web/puzzle/8599447/

eli [12:08 AM]
Just that. I tried one of the whole unmature microRNA and it was boring.
Most of the mir's I tried so far are boring - not sparking much structure.
Our mir was special in that it has 5 A's. The A's seems to help spark extra structure. [Repeat A's and U's seems to help with that by sparking loop structure in Vienna]
But what made me really excited now, is a whole other microRNA.

eli [12:09 AM]

https://www.eternagame.org/web/puzzle/8599478/

eli [12:10 AM]
This one is perfect for a massive switch.
I found it in a paper on alzheimers and microRNA's that are upregulated in alzheimers.
I think it is pretty. I wonder if microRNA's can pair with themselves.
I mean not individually.

omei [12:15 AM]
When you use the word miRNA, are you thinking about the ~22 base segment, or the ~60 base hairpin it is cut from?  (the pre-miRNA.)

eli [12:16 AM]
22 base segment

omei [12:17 AM]
So our last picture is based on ~22 base segment that itself forms a hairpin?
Maybe not -- I'm not sure why I thought that, looking at your picture.

eli [12:20 AM]
uploaded this image: Only when in NuPack


eli [12:20 AM]
But it harbours sequence that is complementary with itself if the sequence is in double.

omei [12:20 AM]
Ok, a relatively week hairpin

eli [12:21 AM]
uploaded this image: image.png


eli [12:21 AM]
I think my surprise stems from that I hadn't expected anything beautiful forming from the microRNA's. As they seem very repetitive.

omei [12:24 AM]
Repetitive in what sense?

eli [12:24 AM]
In that they are biased in sequence. Perhaps lacking one of the base letters.
Or very heavy in U's

omei [12:25 AM]
I see.

eli [12:25 AM]
By the way U's seems to be heavily used in full moving switches.
Example from Let-7 with no C's:

ugagguaguugguuguauagu

omei [12:27 AM]
I suppose one "reason" for this is that they would be less likely to form dimers. (edited)

eli [12:27 AM]
Good point
I like. That could explain.
Also the sequence is cut from what is a hairpin stem and more specifically one strand of it. (edited)
And a strand in a hairpin is different in sequence from a full hairpin stem.

omei [12:29 AM]
But long single strands that form hairpins are also able to form dimers.

eli [12:29 AM]
Plus if the microRNA is too varied in sequence - it will be a good and very static stem. :) Less likely to detatch.

omei [12:30 AM]
... which makes me wonder if pre-mRNAs have anything, like chaperone proteins, to discourage dimers.
Looks like they do require one or more proteins in oder to escape from the nucleus.

eli [12:37 AM]
Is RNA dimers kind of like palindromic RNA

omei [12:37 AM]
Exportin5 looks like it would probably distinguish against the dimer form. https://www.ncbi.nlm.nih.gov/pubmed/14730017
ncbi.nlm.nih.gov
Exportin 5 is a RanGTP-dependent dsRNA-binding protein that mediates nuclear export of pre-miRNAs.  - PubMed - NCBI
RNA. 2004 Feb;10(2):185-91. Research Support, Non-U.S. Gov't
Not sure what you're including in palindromic.  Dimer itself means nothing more than two RNA binding to each other.

eli [12:39 AM]
Ok
Got it

omei [12:40 AM]
Any sequence that simply forms a hairpin can also easily form a dimer with another copy of itself.

eli [12:40 AM]
And since pre-miRNA are the same, they should in principle be able to bind with each other also. I think I get your point
And some miRNA's were also downregulated in alzheimers. I may try make a puzzle of a dimirized preMRNA of one of those.

omei [12:44 AM]
It should work.
Remember in the the MS2 cooperativity puzzles, how the MS2 hairpin sequences tended to pair up against each other rather than forming a hairpin?
... in the absence of the aptamer boost.

eli [12:47 AM]
Yup.
It also seemed to matter what were in between them
In round 1 I placed stems between them. They were not happy.
In round two what worked was them being close together seperated by loop bases, and a weak stem at top.
So a loop is a very different seperator, compared to stems.

omei [12:49 AM]
In retrospect, I think the most cooperativeness might come from putting two directly adjacent.  We now know the energy model incorrectly estimates the energy of adjacent stacks, but we didn't know that then.

eli [12:52 AM]
True. My jumping jack solves (in the RNA input labs) that had a stem between the input and reporter weren't as effective either in the RNA input labs as the solves that had the reporter and the input fairly close to each other and forming a loop between them as they turned off. (edited)
The input and reporter also there benefitted from being close together (coaxial stacked) when both were turned on.

omei [12:55 AM]
As I recall, one of the highest, if not the highest, cooperativity score came with two MS2 hairpins separated by a single base.  We now know that's close enough to get a small coaxial stacking bonus, but not nearly as much as having them adjacent.

eli [12:55 AM]
Ah, I didn't realized that. Well noticed.

omei [12:56 AM]
Unfortunately, I don't remember the specific design.  It was from a player who was not one of the "regulars".

eli [12:56 AM]
I think I looked more of the scores.
As in EternaScores.

omei [12:56 AM]
Right.

eli [12:56 AM]
Yup. I recall there being a new player getting a real fine score
Christopher something
Photo of cynwulf28

cynwulf28

  • 83 Posts
  • 26 Reply Likes
There are a lot of microRNAs out there. I noticed this paper and wonder if it could help to make sense of the miRNA zoo we face: https://www.nature.com/articles/srep25941
Photo of Eli Fisker

Eli Fisker

  • 2224 Posts
  • 486 Reply Likes
I love the image of an RNA zoo, especially that there is even one among the microRNA. Thx for the link. I shall read the paper.
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes
Same sequence - different engine - different structure


Chat with Omei from yesterday:

I have been making microRNA puzzles. I got a Vienna2 switch made. It made me realize that we need more switches done in Nupack and Vienna2.
The vienna2 switch I made from a Vienna switch. [Based on the same sequence]
They were quite different.

omei [8:01 PM]
So player puzzles are tied to a specific engine?  I figured that must be the case, but never verified it.
When another player tries to solve it, is the engine displayed?

eli [8:01 PM]
I did a small illustration with single states:

eli [8:02 PM]
uploaded this image: image.png


eli [8:02 PM]
Some puzzles are exactly the same despite engine.

omei [8:04 PM]
AFK for a few minutes

eli [8:04 PM]
Ok.
Same sequence, different engine:


eli [8:05 PM]
uploaded this image: image.png


eli [8:16 PM]
Cynwulf was hunting a microRNA fold and he switched engine to get a pretty structure. That's what got me inspired.

eli [8:17 PM]
uploaded this image: image.png


eli [8:17 PM]
He ended up using a selfgenerated microRNA to make the above puzzles.
So I instead swapped a switch puzzle between engines.


Seing different structures of the same sequence


To see the different engines in action on the ground state in this puzzle

1) paste in this sequence:

AUAAGACGAGCAAAAAGCUUGUAUAAGACGAGCAAAAAGCUUGUAUAAGACGAGCAAAAAGCUUGUAUAAGACGAGCAAAAAGCUUGUAUAAGACGAGCAAAAAGCUUGUAUAAGACGAGCAAAAAGCUUGUAUAAGACGAGCAAAAAGCUUGUAUAAGACGAGCAAAAAGCUUGUAUAAGACGAGCAAAAAGCUUGUAUAAGACGAGCAAAAAGCUUGUAUAAGACGAGCAAAAAGCUUGUAUAAGACGAGCAAAAAGCUUGUAUAAGACGAGCAAAAAGCUUGUAUAAGACGAGCAAAAAGCUUGUAUAAGACGAGCAAAAAGCUUGUAUAAGACGAGCAAAAAGCUUGUAUAAGACGAGCAAAAAGCUUGUAUAAGACGAGCAAAAAGCUUGU

2) Right click and choose the option Beam to puzzlemaker

(Make sure you don't have the puzzle in multistate view. That will make only one of the structures go to the puzzlemaker in double.)

3) The first state (left) is stable in vienna.

4) Click on the leaf for put the puzzle in Native mod.

5) Switch energy engine by clicking at the engine name where it says Vienna and there are a molecule, at the top left of the screen. Now you can see the alternative structures of the same sequence in different engines.
(Edited)
Photo of cynwulf28

cynwulf28

  • 83 Posts
  • 26 Reply Likes
A word of warning...when you switch your Engines you lose your 'stacks' and will NOT be able to use the undo arrow to a time before you last switched Engines. I have found that out the hard way more than once. 
Photo of cynwulf28

cynwulf28

  • 83 Posts
  • 26 Reply Likes
This might prove relevant: <iframe width="1413" height="672" src="https://www.youtube.com/embed/qZ9cTiqALdw"; frameborder="0" allow="autoplay; encrypted-media" allowfullscreen></iframe>
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes
Got one lab design made and a puzzle up
My current fascination is transposons, but I'm not done with prions either.
I can basically spark a switch as long as I make repeats - almost no matter what starting sequence I use.

omei [9:14 PM]
That answers the question I was about to ask -- whether there seemed to be something unique about transposons or prion mRNA.

eli [9:14 PM]
:)
Sometimes it takes a specific numbers of repeats to spark a structure, like 4 or 8
but some structures only happen if there are an uneven number or rather incomplete extra stretch of one of the starter sequences
I mean sometimes the pretty structure don't turn up if there are 2 repeats
or 4

omei [9:18 PM]
Can you predict what you end up with before you see it?

eli [9:18 PM]
No.
Here is an example that when repeated in 4 is ugly. But if I add an incomplete fifth part, it gets pretty.
Sequence: AUGUUUGCUGGAUUCAAAGAGUUCAAAGUUUGGUCAAUACUUCUUGGAAUGUUUCCAGUAUUUGAGGAAA
CAUGUGACGAUGAUAACGACCUUUAA

(https://www.ncbi.nlm.nih.gov/nuccore/U63707.1)

eli [9:19 PM]
uploaded this image: image.png


eli [9:20 PM]
with a fifth incomplete sequence

eli [9:20 PM]
uploaded this image: image.png

(Made a puzzle of it afterwards: http://www.eternagame.org/web/puzzle/8663375/)

omei [9:21 PM]
Interesting.

eli [9:21 PM]
It happens a lot. Also with 2 repeats and an incomplete third one
So I usually try put in different amounts of repeats to see what they spark of shape
and try fill the whole puzzle also
I can likely make both puzzles switch, but the latter will be more likely to spark symetric switches
Pretty to pretty.
Its hard to go ugly to pretty
But very easy to go from pretty to ugly
Its kind of like entropy. More possible states = more entropy. If there is symmetric order, it is easy to spark more limited symmetric order. But if there is huge disorder, it is harder to go towards lower entropy
Repeat sequence is a way to make order and limit the amount of possible states

omei [9:25 PM]
Makes sense

eli [9:26 PM]
And I also think this is why successful lab switches are so fond of not just repeat sequence but GA and CU rich sequence
Plus likely some physical reasons


Afterthought

I made an earlier puzzle with the exact same sequence, where I could get exactly 4 copies of the sequence to spark a beautiful structure as is and make a full symmetric switch.


http://www.eternagame.org/game/puzzle/8658067/

Something I found much harder to make with the puzzle I made with 4 "1⁄2" sequences. I couldn't make it fully symmetrically either in the second state.

The only difference between these two puzzles are that in the first I had inverted the sequence - to make it identical to the real life mRNA. So this is just another example of direction matters. :)

It is not as if nature designed these sequences with the thought that I would come along and just for the fun of it make switches of them. So I wonder about the why. Now I can't help wonder if this is generally the case for the designs that seems to need to get an extra part of the sequence along to make pretty structures, that they are of the uninverted kind?
(Edited)
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes
RNA binding proteins binds to repeat RNA base sequences


Rhiju shared a super interesting paper. He described it as the following: an interesting paper on which RNA sequences are most likely to be bound by proteins in human cells.

Sequence, Structure, and Context Preferences of Human RNA Binding Proteins

He shared this image along. 


Here is the main idea of the paper: 

"We find that many RBPs (RNA binding proteins) bind a relatively small, defined subset of primary RNA sequence space that is rich in low-complexity motifs composed primarily of just one or two base types." 

Rhiju: When I saw these sequences, I thought of some of your points about recurring sequence motifs (like magnet regions) in switches. Maybe biological proteins have evolved to toggle natural riboswitches inside our cells?

Eli: So you are thinking bigger proteins, not just smaller molecules or peptides. Perhaps this may explain why there is also a high ratio of repeat bases outside of the aptamer region in natural occurring riboswitches. In the "static" region.

The paper said they used several fixed aptamer sequences and mutated the sequence around them. I can't seem to find out which aptamers they used. Anyway, from what I have seen so far, pretty much all aptamers carries base repeats and magnet segments. And so do their surrounding sequence.

Ok, so basically you are guessing that perhaps proteins can also interact with aptamers in our body, not just the molecule the aptamer is supposed to bind, right?

Rhiju: yes! actually these little sequences appear to act as aptamers for the *proteins* not small molecules.

Eli: Ah. :)
But it is the same principle. Just put in place two different places.
Lol, protein aptamers. :)
So RNA aptamers have repeat bases and sequence bias. Despite they are to bind with a wealth of different kind of molecules. RNA binding proteins seems to favor binding to RNA with repeat sequence and RNA with sequence bias.

rhiju: yes, it is indeed the same principle!

eli: I wonder what physical reason there can lay behind this?

rhiju: yes i wonder too
this is a frontier area in RNA biology
and in RNA medicine

eli: 
I mean I suspect that repeat bases let easier go.

rhiju: 
essentially every RNA in our bodies is thought to be bound (maybe even 'coated') by proteins
but we have few computational tools for modeling these RNP complexes
and even fewer ones for designing them

eli: 
Because I observed that for static designs, that if we used several GC pairs in a hairpin stem and if we didn't cross (flip) at least some of them, then they would easier slide or mispair somewhere else.

(Tutorial: Crossed GC pairs in stems #1 - High crossing frequency)

rhiju: but that understanding will be crucial for designing therapies that, e.g., silence RNA genes

eli: The variation seemed critical to stability
Whereas well mixed bases seem detrimental to switches.

rhiju: exactly

eli: So I think something special happens when there are repeat bases. Something that will allow for easy bind but not too strong a bind. I recall I mentioned base size. That pyrimidines were short and purines being long. And having long bases side by side, and short bases side by side, made for an easier split. 

Background: Why do switch RNA like repeat sequence?





Idea for comparing eterna switch sequences with protein binding RNA sequences


Rhiju suggested that we could take a look at which of the switching sequences in the best eterna switches that match the list in the paper with the protein RNA binding sequences. 

(Edited)
Photo of rhiju

rhiju, Researcher

  • 403 Posts
  • 123 Reply Likes
Thanks, Eli! What would help would be a bioinformatic analysis of some switch solutions (even in silico switches) to see which sequences are over-represented in the switch regions. Perhaps another player with some coding skills could help out?
Photo of rhiju

rhiju, Researcher

  • 403 Posts
  • 123 Reply Likes
Also, one more comment -- Nando has written the best bot at solving singe-state puzzles, called NEMO (https://www.biorxiv.org/content/biorxiv/early/2018/06/14/345587.full.pdf).  It requires having some approximate rules for what nucleotides to put in loops & stems.

Perhaps we could help Nando to make a NEMO for switches if we could collectively compile nucleotides (or stretches of nucleotides?) to guess for switch regions.
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes
In silico switches versus lab switches

Rhiju, you mentioned in silico switches also. 

There is one additional thing we need to take into account when it comes to in silico switches versus lab switches. 

Just as single state in silico puzzles are not the same as single state lab designs. In silico switches and lab switches are not entirely the same animal. Things that are legal for puzzles, will be illegal in lab. 

What will be an optimal base frequency for the switching areas in a switch will strongly depend on what shape the puzzle has. 


Asymmetric switches versus symmetric switches

An asymmetric switch - like most of Malcolm's switch puzzles - will need an entirely different base frequency compared to symmetric switches - like many of Cynwulfs. 

  • The symmetric switches demands heavy base repeat - be they in silico or in lab. 
  • The asymmetric switches like Malcolm's demand a solving style that looks more like a static design - as in more mixed bases. So AU base pairs like AUAUAU. (This is still a very repetitive sequence pattern though - it is the sequence that is repetitive and less the individual bases.) And some of their GC pairs in stems crossed. Which is also typical of static RNA designs.

I noticed this discrepancy between switch lab and switch puzzles, since I tried solve Malcolm's puzzles in the style I had come to expect to work for lab switches with repeat bases and his switches puzzles utterly resisted my attempt. 

So what will be optimal base solving frequency for switches will heavily depend on the structure of that switch. 

I think we will find that just as there are static RNA shapes that are impossible to make stable in lab (the symmetric ones - if the puzzle is not very small) despite them being possible in silico. 

I suspect we are also going to to find that there are switch shapes that are impossible to make make switch in lab. (potentially the more asymmetric ones)


Position of switch elements in lab switches versus switch puzzles

One more thing that is done in switch puzzles, that may not be possible in lab: 

A lot of the puzzle switches that eg hold the FMN aptamer has the FMN aptamer very far away from the rest of the puzzle that it is going to interact with. In lab switches the switch elements seems to favor particular positions and a certain distance range in relation to each other. Also depending on if it is an ON switch or an OFF switch. 
(Edited)
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes
Hmm, perhaps this is the explanation on something I have been wondering about.

I had been wondering about the strong repeat base frequency in certain past static RNA designs that structurally were heavy in symmetry. 

So far I had explained this oddity in static designs with that long stems are more way more tolerant to any kind of pattern, repeat bases too, than shorter stems. 

I still think this is true. 

However there may be an extra reason. That heavily symmetric puzzles that are supposed to be static, may somehow benefit from repeat bases. Just as heavily asymmetric switch puzzles seems to benefit from well mixed bases. 

Judging from the amount of repeat bases and repeat CU and GA in particular this sequence ought to be a switch. 



https://eternagame.org/game/browse/24153/?filter1=Id&filter1_arg2=25205&filter1_arg1=25205

When it is not it may be because of how these repeats are used. They are made so the sections individually are making it only possible for neighbouring strands to pair with the intended neighbour and not the neighbour at the other side. 



This design was the top scoring design in The Cross - pre cloud lab. 

The design was still a winner in the cloud lab rerun. But other sequences were better solves. Some of the base repeat sequence disappeared in the new winners. 


Sum up of what a RNA design really wants

Since switch switch designs seems follow the opposite rules as those for static designs, so I could sum it up like this:
  • symmetric switches want repeat bases and biased sequence - also in stems
  • asymmetric switches want more mixed bases with some repeat sequence
  • asymmetric static designs want well mixed bases in stems and some repeat bases - mainly in loops. 
  • symmetric static designs want well mixed bases with a lean towards more repeat bases - also in stems
While I think symmetric switch RNA and asymmetric static RNA are the ideal. 
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes
Downboosting around stems in switch multiloops 


One of Cynwulfs new puzzles caught my attention. 

This puzzle had only two areas left to solve. The static stems outside of the switching area and a small part inside of the switching area. It was the small part inside the switching area that specifically caught my attention. 

I have highlighted (black rings) the downboosting areas - bases that raise energy and make it more positive next to the stem they are placed around. 

The more switching an area is involved in, the more downboosting seems to happen in the switching multiloops. 


State 2


State 1


For lab switches there has long been a trend that the bases just around a switching stem - either at the bottom of the stem or for the two boost spaces in the loop, trends toward being downboosted. Having the bases that is added causing a more positive energy. 


Positive switch multiloops/Negative static multiloops?

As I have earlier observed the RNA folding rules for switch RNA are the opposite to those for static RNA. 

Basically if it works for static designs, it is worth disrupting in switch designs. If it works for switches, it is worth trying to disrupt for static designs. I have seen it again and again. 

For static designs, boosts that made energy more positive around stems in multiloops that are supposed to be static were detrimental to good stable RNA designs. This is one of the areas where the RNA algorithms failed big, because they didn't seem to know. Vienna1 in particular.

I have noticed that for labs that making energy more positive for a multiloop (destabilizing) seemed to help in switch labs. However I haven't seen downboosting for multiloops go on to the same degree yet as in Cynwulfs puzzle. 

With energy turned on, extremely positive loop and multiloop energy: 



Static designs favors higher negative energy inside multiloops - although it can get too negative also. Switch designs tends to favor higher positive energy inside multiloops in the switching areas. Although it can probably also get positive too. 


Perspective

I wish to see what Cynwulf did with downboosting multiloops in a switch puzzle, tried in lab. I will be very interested in if the puzzle pattern with multiloop downboosting can be transferred to lab.  



Background posts: 

I have given an introduction to downboosting here: 
Quick overview of the incoming results
Downboosting hairpin loops etc <-- See the section or search for downboosting.


Multiloops - Static designs
Blue, green and red nucleotides in multiloop ring
Clean halo in multiloop ring or flaws of Vienna
Something loopy – energy in multiloops


Multiloops - Switch designs
Making a switch happen by weakening multiloop in unbound shape (in the switching area)



(Edited)