Player/dev collaboration needed to create an effective, rapid, in-round Select-Test-Publish-Analyze cycle.

  • 7
  • Problem
  • Updated 4 months ago
  • (Edited)
There are a lot of steps needed to transform our success in OpenTB Round 2 into a working diagnostic. The Gates foundation has given the Das Lab a grant for the "hardware" part of that transformation.  But when the dev team started thinking in more detail about the whole process, we realized the weakest link in the current process is getting rapid player feedback, in terms of new designs that address changing requirements, as the engineer/experimentalists narrow down the detailed design of a point-of-care device.

The most obvious case-in-point of this is getting array results for the Round 4 OpenTB designs. This has been an issue because Johan Andreasson, the post-doc who helped pioneer this technique and who personally conducted the Eterna array experiments, has gone on to a new job. He is still available for consultation, but can't spend the time needed to guide the experiment through its many steps. Fortunately, Feriel Melaine, the new post-doc who has successfully replicated the array results for the AK2.5 design, but with a bead-based experiment, has now agreed to take on the task of getting us array-based results for the Round 4 designs.

But in addition to that, we realized that we didn't really have adequate structure in place to do a rapid test cycle on the order of one every two weeks. The lab believes they can get their part of the work (receive the list of designs, order and receive the DNA templates needed as inputs to the experiments, run the experiments and return the data to players) in one week.  But a two week cycle implies that players would then have only one week to look at the results, analyze them, disseminate that analysis to other players to create and submit new versions of their designs and then collectively decide on which designs should be submitted to start the next rapid feedback cycle. We have no precedent for having done that before.  Now, we need to create a process to do that, and we'll be using the extended Light-up Sensors project as our testing ground.

Hoping to help organize the discussion a bit, I propose subdividing the discussion of the player part of the process into some individual steps. Starting with the next step falling into players' laps with the first 6 designs (currently in the lab):
  1. Players (typically the more experienced ones) analyze the results and try to distill what it "means".
  2. Analyses are disseminated in a form that is easily accessible by all lab players.
  3. Players submit new designs based on what they think they have learned from the analysis and subsequent discussion.
  4. Players select a new set of designs for the next rapid feedback cycle.
I'll immediately follow up with a separate "Reply" on each of these substeps, and if that organization fits with what you want to say, you can "Comment" on the corresponding "Reply".  But there's no need to feel constrained to that structure if you choose.
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 960 Posts
  • 302 Reply Likes

Posted 8 months ago

  • 7
Photo of rhiju

rhiju, Researcher

  • 403 Posts
  • 122 Reply Likes
@omei, nice post in game! Time to fill out the placeholder text in this getsat post. =)
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 960 Posts
  • 302 Reply Likes
Analysis step

What techniques or tools do you useful in analyzing results?  What can we do to make it easier for other players to use them?
(Edited)
Photo of whbob

whbob

  • 182 Posts
  • 56 Reply Likes
In past labs, I like to go to the classic data browser, select a design ID and use the U and D keys to step through the 2D images. Sometimes players place their unique label for their design in an area of title or comments that gets cut off in the list view. When that happens, there is no way to tell what the designID is for that image. Is there a way to have the designID come up in the 2D image of a design?
Photo of whbob

whbob

  • 182 Posts
  • 56 Reply Likes
In past labs, I like to go to the classic data browser, select a design ID and use the U and D keys to step through the 2D images. Sometimes players place their unique label for their design in an area of title or comments that gets cut off in the list view. When that happens, there is no way to tell what the designID is for that image. Is there a way to have the designID come up in the 2D image of a design?
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 960 Posts
  • 302 Reply Likes
@Zama It's my feeling that Vienna2 has been making the best predictions, but it is not a hypothesis I have tried to rigorously test.  That would be a great project for someone to take on.  It wouldn't have to be a  huge undertaking.  Maybe just start with a list of the designs from a recent puzzle, along with their scores, and then go through and check off whether each folding engine predicts it will switch or not.  I could help with producing the starting spreadsheet of designs and then with the summarization.

The main reason for focusing on Vienna2, though, is that it is the version I am most familiar with using in a batch processing mode.
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 960 Posts
  • 302 Reply Likes
@whbob That's a very good suggestion. At the moment, we're focusing on getting rid of flash while not changing the game functionality, so that we have a clear comparison between the two.  But I'll see if this is something small enough that we can slip it in somehow.  Ah! With a booster. :-)
Photo of flagchina

flagchina

  • 1 Post
  • 0 Reply Likes
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 960 Posts
  • 302 Reply Likes
Dissemination step

Standard practice would be to create a forum topic for the rapid feedback round and not try to further organize the discussion.

My impression, though, is that most players who are submitting designs do not routinely follow a forum thread, if they ever see it at all.  There must be a better way to reach more players. But what is it?
(Edited)
Photo of jandersonlee

jandersonlee

  • 543 Posts
  • 120 Reply Likes
The comments field for designs could possibly be used after the lab for annotations regarding a design. The votes field could be used as a like function. That way people could comment on designs they think might be worth of study/mutation. Having a way to rank the designs by number of likes and/or comments would make it easier to find this information.
Photo of DigitalEmbrace

DigitalEmbrace

  • 18 Posts
  • 13 Reply Likes
For me, a document that briefly outlines the current best practices (or results analysis) for designing in a lab would be helpful. I also think a glossary would be helpful for quick reference.

For instance, a few days ago mgotrik popped onto chat to make sure players knew to avoid entangled aptamers in the lightning rounds. 1) That info could be added to a best practice doc. 2) "Entangled aptamer" is one of the terms not clearly defined for me. I spent an hour reading through forums and am still not completely sure. (Is an entangled aptamer the same as sell-turnoff aptamers? Or does it imply bends in the aptamers also?)

Another example, I'm not sure all players are aware that Vienna2 is delivering the most accurate model in the lighting rounds. (Or is this not true?)

I guess the best way to implement either of these suggestions would be a google doc linked in the main lab page or in one of the News posts. The google docs could be constantly updated by advanced players as we discover better practices and invent new terms. I'd be happy to help edit any docs for clarity and brevity.
Photo of jandersonlee

jandersonlee

  • 543 Posts
  • 120 Reply Likes
Vienna2 is not necessarily the best model, it's just the one for which we have the more detailed tools (arcplot). In theory we could add NuPACK to arcplot if someone had the time to do a little coding for me.
Photo of Eli Fisker

Eli Fisker

  • 2216 Posts
  • 477 Reply Likes
Hi DigitalEmbrace!

An entangled aptamer is two different aptamers that are entangled in such a way that for the one to form, the other have to be literally broken. So the aptamers are taking turns excluding each other. It's a trick that Jieux started in the FMN exclusion lab. It can only be used for exclusion designs. 



Here is a tutorial puzzle introducing the concept. 

https://eternagame.org/web/puzzle/8827766/

An aptamer selfturnoff is different to entangled aptamers. 

AN aptamer selfturnoff is the sequence of one aptamer, that when the aptamer is to be turned of, is turned off within itself. 

Here is a scripted tutorial demonstrating this in practice. 

https://eternagame.org/web/puzzle/8957760/
(Edited)
Photo of DigitalEmbrace

DigitalEmbrace

  • 18 Posts
  • 13 Reply Likes
Ignore what I said about Vienna2, I can't remember where I got that idea. And I just found the dictionary on the Wiki. Doesn't cover newer terms but some good info for me to review. You guys have put so much time into creating tutorials and guides for new players, really impressive!
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 960 Posts
  • 302 Reply Likes
Design step 

What techniques or tools do you use to analyze results?  Is there something we can do to make it easier for other players to learn about and/or use them?
(Edited)
Photo of jandersonlee

jandersonlee

  • 543 Posts
  • 120 Reply Likes
I tend to use the lab tool a lot, plus some of my own javascript-based tools. A first step is usually to look at better scoring designs and pick out a few that are sufficiently "different" from each other to warrant parallel mutation -- i.e. not all eggs in one basket. (I don't tend to do much original design these days.) I use the Mutation/Submission Booster a lot for this step, typically after manual edits to strengthen necks/stacks aided by the dot plot data. I'd like to be able to prune not just on whether or not it passes a design engine, but on the delta between the MFE in the two states. What might also be nice is if players could vote on designs as potential targets for mutation. This might help less experienced players get a hint on what to try to modify and could be useful if you thing there are more potential targets than what you can manage on your own.
Photo of DigitalEmbrace

DigitalEmbrace

  • 18 Posts
  • 13 Reply Likes
I use the mutation booster regularly to look for new designs. After pruning, I click through designs to find desired fold shapes. I don't submit all mutations, only the ones I like. And Jeff's new arcplot tool, holy cow, blowing my mind. I use it not only to identify good designs, but also to find and fix weak pairings.
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 960 Posts
  • 302 Reply Likes
Selection step

One thought here is to create a way to add automated bot analysis to player analysis.  We could simplify the creation of evaluation bots, so that players need only to express the evaluation function into code. For example, it would be reasonably easy for the dev team to write a generic evaluation bot that a player could tap into with a minimum of scripting skill:

Suppose you thought that having a specific  range of GC pairs in each state was important for a good switch.  You might write an EternaScript as simple as this.
function Evaluate( design ) {  
if (design.state1.GC_percentage >= 50
and design.state1.GC_percentage <= 70
and design.state2.GC_percentage >= 40
and design.state2.GC_percentage <= 60)
score = 100;
} else {
score = 0;
}
return score;
}

(Of course, advanced scriptos could write a much more complex evaluation function.)

You would then tell the player-led development team the ID of your script, and it would be added to the list of evaluation bots.  On a regular basis, the team would update a spreadsheet and/or fusion table that evaluated all submitted designs with all bots.  Players could use these predictions to inform them in selecting designs for the next rapid feedback cycle.
(Edited)
Photo of jandersonlee

jandersonlee

  • 543 Posts
  • 120 Reply Likes
Although we may ultimately find some metrics that work, I'm not yet convinced in this regard, still it may be a fruitful line for some to pursue. What might be interesting though is for some of the "bot" designers (e.g. ViennaUCT, SaraBot, ...) to express what metrics *they* use in a way that might help players to do similar manual or assisted design.
Photo of rhiju

rhiju, Researcher

  • 403 Posts
  • 122 Reply Likes
perhaps we could agree to all use one spreadsheet or fusion table, and individual scripters could add in columns based on the outputs of their favorite bots.
Photo of JR

JR

  • 235 Posts
  • 17 Reply Likes
I would be interested in the spreadsheet type summary contribution of what players can conclude from say, the top 12 designs. A sort of what characteristics do they think contributed to this design succeeding. Then players can interpret the results back into their individual designs or their bots. If you want to get snotty - create lab periods - No contribution, no lab/puzzle points for that period unless contribution to spreadsheet, but I think that would be too harsh, probably should just go for an analysis badge.
Photo of Gerry Smith

Gerry Smith

  • 55 Posts
  • 25 Reply Likes
Given the visual analytics eterna thrives on, is it much of a leap for the visual gaming programmers who are making such inroads at developing programs to beat computer games to come up with script-like visual tools that players could use to test strategies and find designs with the parameters they set? How to expand eterna into this community?
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 960 Posts
  • 302 Reply Likes
The data and presentation is still preliminary, but I want to let players know that the first rapid feedback, bead-based experiments were a success!

Of the six design, one for each puzzle was verified to have a significant fold change.
 
What we have learned so far from this process falls in several of the four sub-replies I originally broke this forum topic into.  I'm going to add additional comments in the most relevant reply threads.
(Edited)
Photo of Eli Fisker

Eli Fisker

  • 2216 Posts
  • 477 Reply Likes
This is super interesting!

These fold change graphs seems have predictive power, just as the Wuami graphs we got for labs with inputs earlier. 

Could we have a similar solution to what we got then?

  • A server that can give us immediate feedback on a design while we are designing? 
  • Or a fusion table over the already computed active lab designs?
I find it interesting in the above image that there seem to be a trend for that the raw data give a downward moving trend at the end (right) where as the real life data tends to give a more raising curve or a slower dip at the end. 
Photo of Eli Fisker

Eli Fisker

  • 2216 Posts
  • 477 Reply Likes
Entangled aptamers

My design with with entangled aptamers that I am curious about, didn't work as intended. However I wish to highlight that Zama's design which did show fold change, indeed has entangled aptamers. 

              

I will attempt to move the two entangled aptamers both closer but also further appart. I expect this will affect the fold change either for better or worse. I would really love to be able to run the comming Zama mods through a fold change estimate server. :)

Photo of Zama

Zama

  • 33 Posts
  • 9 Reply Likes
And Zama would like to see them too. :)
(Edited)
Photo of Astromon

Astromon

  • 176 Posts
  • 22 Reply Likes
I didn't realize  Zamas fmn puzzle was rapid tested I thought it was the trytophan? did
this fmn get tested also?
Photo of Zama

Zama

  • 33 Posts
  • 9 Reply Likes
Oops, you're right Astromon- it was the Trypto B-SS ZZ-1-18 that was rapid tested, not the FMN  A- Exc. ZZ-1-18. I guess I need to find a better numbering system. I totally missed that- good catch!
Photo of Eli Fisker

Eli Fisker

  • 2216 Posts
  • 477 Reply Likes
Sigh! Astromon and Zama, you are both right. I went for the wrong FMN lab. 

When I now look at the correct design switching design by Zama, it is of the type that has aptamer selfturnoff instead. Which is a nice result also. 
Photo of Eli Fisker

Eli Fisker

  • 2216 Posts
  • 477 Reply Likes
Astromon had a question on the new fold change graphs


Astromon: on another note what did you see in those graphs?

eli: graphs - like the fold change ones?

Astromon: ya

eli: I see the future... :)

Astromon: was wondering about the raw data part

haha

eli: I suspect that the raw data part are like simulation. Calculated in energy model before the designs are run in lab.

Astromon: wow ok!

eli: So they are like prediction points. What is so promising is that the prediction points are so close to the actual data.

Astromon: indeed

have seperation

eli: Which is why the faster we get a server/tool/script that can make these graphs, the better designs we can make

Astromon: yay!!!

they working in Vienna2 predictions

eli: So these new graphs are akin to the Wuami charts we got for the TB labs. They have predictive power.

Astromon: sweet

eli: They can give us a sense for if we have our design switch in the right direction and give us fast feedback on if what we are playing with have a shot at working.


Photo of Brourd

Brourd

  • 433 Posts
  • 79 Reply Likes
Perhaps I'm just misinterpreting the origin of those graphs that Omei posted in this forum post, but I believe they are the graphs with the experimental data that Michael in the Das Lab generated from the new experimental procedure. They are functionally equivalent to the graphs that Johan produced through the custom Illumina sequencer/microscope, just without the fancy stuff.

One possible origin for the "raw data points" and the decrease in response as a function of concentration could be that they are raw data before a statistical correction. Response may decrease as a function of time, whether it's due to degradation of the RNA, or possibly some kind of change in binding affinity as malachite green increases in concentration.
(Edited)
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 960 Posts
  • 302 Reply Likes
I'll try to provide more specific details as I learn them, but basically, I think Brourd's description is accurate.
Photo of Eli Fisker

Eli Fisker

  • 2216 Posts
  • 477 Reply Likes
Sigh. I wish I had been right. 
Photo of Zama

Zama

  • 33 Posts
  • 9 Reply Likes
Is there a better way of picking designs to test then voting? It would be nice if each player could choose their favorite design for testing. If you can't test that many then allow the players with the most designs pick until which time you can do one for each. Or some variation of the above...
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 960 Posts
  • 302 Reply Likes
Hi Zama! Just in case it wasn't clear, all the designs will be tested at the end of the round, as before. It's just this fast-track process for (limited) rapid feedback that we're working an adding as a new component to the process.

Player voting has a lot of advantages, first and foremost the involvement of players. What I think we need, and am working on providing, is more immediate evaluation of designs, using multiple criteria, so that players can focus their human evaluation on the relatively few designs that fall in their category of interest. We're a long way from providing that now, but I really think it will spur progress.

This is not to say that we are absolutely stuck on using player voting as the final decision process, so keep up the discussion.
Photo of Zama

Zama

  • 33 Posts
  • 9 Reply Likes
No problem, Omei, as always, you were clear. But for me, these are the first test results I've ever seen since starting to play Eterna. I have long forgotten what I might have been trying to accomplish in those previous labs. I thought it would be nice for everyone to get a chance to glimpse the efficacy of their master plan- before they forget what it was ;)
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 960 Posts
  • 302 Reply Likes
I am creating a Google doc with arc graphs for as many voted-for designs in the two current fast-track puzzles as I have time to do.  Currently, it has the five designs that were at the top of the Theophylline B Same State (MGA) designs, sorted by vote.  Here's an example
I'll switch to Tetracycline B Same State (MGA), do five there, and then see where my day stands.
Photo of Poll na gColm

Poll na gColm

  • 26 Posts
  • 20 Reply Likes
Is this an error? C-C pair and A-C pair and no energy details?

Photo of Eli Fisker

Eli Fisker

  • 2216 Posts
  • 477 Reply Likes
Hi Poll na gColm!

This can happen sometimes. I am not sure why. However I found a sequence of yours that is closely related to the one you show above.  

I can reproduce your output with it. 

    

Link to the design: 
https://eternagame.org/game/browse/8791823/?filter1_arg1=8986067&filter1=Id&filter1_arg2=898...

Sequence: 
AAUAAUGUUGGAGGAUAUGUGCGGUAACGAAUGCCUUAAUUCUGGUAACGAAUGGCGAAACGCCCGACAGAAGGCCGACGCACAA

P.S. Try swap to the States mode, it will give you a slightly less pretty but more interesting view of your design. 

      

I noticed that the aptamer in the arc plot didn't turn up the same place in your design. Your design is a double aptamer design. The tool goes for the strongest of the aptamer and draws its blue lines to highlight that one. 

I have highlighted the two aptamers in your puzzle. 

    

See the section in the Arc-Plot Tool Intro called Marking and Automarking.

Photo of jandersonlee

jandersonlee

  • 543 Posts
  • 120 Reply Likes
Ah, yes. It is a Double Reporter (MGA) case. The tool is supposed to detect multiple potential reporter sites and pick the one with the better energy, but in this case it seems to be fooled. I guess multiple aptamer/reporter sites needs more work. Any two bases can be marked as a pair. If the PAIRING% is too low, it treats is as zero and omits the details.
Photo of JR

JR

  • 235 Posts
  • 17 Reply Likes
The Arc Plot is a nice tool. Might create a one line output of stats on bottom so we can cut/paste into comments in the lab that is easily readable.. Thanks for your efforts.   
Photo of jandersonlee

jandersonlee

  • 543 Posts
  • 120 Reply Likes
What information would you like in a one-line summary? I've been grabbing all four lines myself at least for descriptions. You could add each line as a separate comment.
Photo of JR

JR

  • 235 Posts
  • 17 Reply Likes
might just html it so it copies into 4 lines. My copy now displays one single line which makes it less readable. Yes, I know, I am pretty lazy, I could do that myself. 
Photo of JR

JR

  • 235 Posts
  • 17 Reply Likes
Is the Arc Plot tool using Vienna2. I have Nupack designs that "work"  where Vienna2 says not working. The Arc Plot data seems to support Vienna2 version of the designs. Is this correct?

Photo of Eli Fisker

Eli Fisker

  • 2216 Posts
  • 477 Reply Likes
ARC-PLOT PATTERN IN WINNING DESIGNS

What do these winners have in common?


Here are 4 winning designs all from different labs, but with the best fold change of their lab. (I have ignored any design that has a fold change error above 1.25)


  • Anything stands out to you?


  • Just focus on the green lines. Ignore all the black and grey.

  • Now the big outer circle is the neck (static stem). Smaller green circles are static stems in the design. Also ignore these.


Zama asked: “Are you referring to the S wave?”


You are getting there.


  • What can be done with the S wave? Could you highlight it in an image?

  • Just pick one of the arc plots above and draw the S wave

  • What would it look like if you follow the line of the green shape?


  • Now do anything stand out?

  • What do these designs have in common?

  • You can draw a line through the green arc plot part of the switching bases.


The shape of the arc plots can take many forms, also for winning designs. Even designs with messier arc-plot shapes can be winners. The above ones are from different labs. 3 Same State and 1 Exclusion design. 


Thx to Zama for being my guinea pig for how I could share this story.



Link to the designs


https://eternagame.org/game/browse/6369184/?filter1_arg1=6456004&filter1=Id&filter1_arg2=6456004

https://eternagame.org/game/browse/7568412/?filter1_arg1=7650114&filter1=Id&filter1_arg2=7650114

https://eternagame.org/game/browse/7559901/?filter1_arg1=7721394&filter1=Id&filter1_arg2=7721394

https://eternagame.org/game/browse/6369186/?filter1=Id&filter1_arg2=6423405&filter1_arg1=6423405



Arc plot of a low scoring design


Notice that many of the green tracks can’t be connected between the top half of the arc-plot (state 1) and the bottom half of the arc-plot (state 2)


Score 32%

https://eternagame.org/game/browse/6369196/?filter1=Id&filter1_arg2=6487563&filter1_arg1=6487563




(Edited)
Photo of Eli Fisker

Eli Fisker

  • 2216 Posts
  • 477 Reply Likes

THE MAZE OF THE ARC-PLOT


I have been drawing on Omei's EternaCon presentation slide. Notice that the bot designs takes much longer roads.



Observations so far


Jeff has made an awesome new RNA visualization tool. The Arc Plot Tool. I have focused this new lab microscope at the usual suspects - past lab winners. Here are some tendencies. 


  • One can draw one road through the arcs in the arc plots of a good design.

  • The shorter the road, the better the switch

  • The road the arc-plot take in human designs is typically shorter than the bots. (Rhiju calls the robot arc plots swirly)  

  • The road through the arc plot takes different types of routes.

  • There is a high fraction of the bases that are directly involved in switching, that switches in both states (bound in both states, but somewhere different)

  • Magnet segments runs through the arc plot roads



One road track


The winning designs I have checked so far - even if structurally very different - seems to have something particular in common.


One can draw one road through the arcs in the arc plot. With the focus only on the part of the sequence that do base pairing and who's base pairing in state 1, also becomes base pairing in state 2. Then follow the route where base pairing connect between state 1 and 2. So where arcs are touching between state 1 and 2.


What this means in short is that bases that are bound in state 1 also a high likelihood of being bound in state 2 (but somewhere different).


Example design with the best fold change for Exclusion NG 2, Round 101. Score 96%, fold change 37.35. This is a blueprint style design with a circularized arc plot

https://eternagame.org/game/browse/6369184/?filter1_arg1=6456004&filter1=Id&filter1_arg2=6456004


I marked the bound and unbound bases in the design above. What the arc road means in practice is that bases that are bound in state 1 also a high likelihood of being bound in state 2 (but somewhere different). Which is particular handsome to know when one is to design from scratch.


Predictive power and a strong tool for designing


If one know the position of the reporter and the aptamer, one can predict very likely rutes a design will need to take - to make it get a race track arc plot.


Take PWKR’s FMN/MS2 double aptamer design. (Winning mod of it) One can literally draw up the known part - the position at where the MS2 and double FMN is placed in state state 2 when they are to bind their molecules. Then knowing the race track benefit, one can draw up the missing part. There will be only few options left.



Then one could draw the road through, like this:



I actually did the above drawing first, based on the winning double aptamer design. After which I realized that I could have guessed the road, had I not known how to solve the double aptamer in the first place.



Omei's different race tracks


Another thing I'm also interested in is that this one way road through the arc plot takes different roads depending on the design solve style. Omei calls the roads for race tracks. He began wondering how many unique racetrack patterns there could be with six crossing points. He drew up classes of them.


Omei's Unique racetracks with six crossing points

I for the fun of it came up with some names based on the shapes.


Class 1: E type

Class 2: Snake type

Class 3: Spiral type (Fibonacci)



Magnet segments


I have been talking a lot about magnet segments earlier. I have also drawn how they interconnected between states. 


Lab overview drawings


However it shows up much clearer in arc plots. Here it can be followed how the GG and CC cross between the states being connected for the whole or part of the way. They turn up like this in a lot of the other winners too. There typically are one such double GG and CC around. Here they are in the center of the switching strand of bases.




Racetrack width


The road isn’t always straight when it crosses over from state 1 to state 2 or in reverse.  


It is sometimes skewed. I think the skewing can be beneficial too, I think the skewing can go to help the bindings unstick easier.


There is a specific spot where the race track meet between state 1 and 2, that is more skewed apart from each other than other places. One place it often happens is at the neck spot in a design.


https://eternagame.org/game/browse/6369184/?filter1_arg1=6456004&filter1=Id&filter1_arg2=6456004


I have noticed this in some of the riboswitch designs like the above one earlier. There are overhangs between the regions that pairs. I tried eliminate those for a period because I found a clean switch between state 1 and state two switching strands would be pretty. Like all switching bases in 1 state become switching bases in state 2 also and no extras. Plus opposite. 





(Edited)
Photo of Eli Fisker

Eli Fisker

  • 2216 Posts
  • 477 Reply Likes
Cynwulf, this is a good question. I agree with you that this is a most mysterious arc-plot. It looks most like a static RNA design. 

In States

In Natural: 


I found your design in the lab: https://eternagame.org/game/browse/8787266/?filter1=Id&filter1_arg2=8885846&filter1_arg1=888...

One thing that may help explain the arc-plot is that the design is not stable in vienna2, while it is in Vienna and Nupack.

 The tool can be used with Nupack and Vienna too, but it may explain part of the weirdness in the above design that it is not stable in Vienna2. To be honest I don't know. Jandersonlee is best in this department. 

There is a section on energy models in the Arc-Plot Tool Intro

There are other strange arc-plots. Mixtures. Show us your weirdest. :)


 

Photo of jandersonlee

jandersonlee

  • 543 Posts
  • 120 Reply Likes
The arcplot tool may eventually be able to use other energy models once it is a Booster. In the meantime, designs that only work in other than Vienna2 may look weird, at least in Natural mode.
(Edited)
Photo of DigitalEmbrace

DigitalEmbrace

  • 18 Posts
  • 13 Reply Likes
I was running through the arcplots of a mutation series yesterday and noticed several designs had a clean arcplot and good numbers (percentages/ratios/kcal) while several other designs had a noisier arcplot but very high numbers (99% 300x). It seems the current thinking is that a clean arcplot with a racetrack is a better indicator of successful switching than high numbers?
(Edited)
Photo of jandersonlee

jandersonlee

  • 543 Posts
  • 120 Reply Likes
We don't yet know what is best, so I'm trying to submit an across the board variety so that we can see what works. That said, my sense is you don't want the dotplot/arcplot "too clean" or it may indicate a reluctance to switch (a few highly stable states) nor too dirty (largely unstable). For this round at least I'm throwing in designs from 5x to 5000x, but mostly 20x to 500x as long as the ON percentage is over 50%. Also there are several dependency chain patterns (loop, spiral, double spiral, mixed) so I'm also going for a variety of those.
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 960 Posts
  • 302 Reply Likes
@DigitalEmbrace The motivation for liking a "clean" arcplot is that the folding engine predicts that there won't be a lot of chance for the design to fold into ways that are neither of the desired ON and OFF foldings.  My own observations of past data lead me to say that it is a good criterion for avoiding many "mushy" designs, but by itself is certainly not a definitive indication of what the measured fold change will be. If anybody had already figured out a good way to predict the fold change, there would be no need for us to be doing these experiments.  
Photo of whbob

whbob

  • 182 Posts
  • 56 Reply Likes
Looking at Round 3 votes now (3:46 pm EDT) ... there are none.  Voting must have ended early?
Are the round 1,2 & 3 fold change graphs going to be available is a particular location? In the wiki perhaps? 
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 960 Posts
  • 302 Reply Likes
There must be some confusion about which puzzles are being drawn from.  They are FMN B Exclusion (MGA) and Tryptophan B Same State (MGA).  These are the same two puzzles we drew from in Round 1.  Make sure you sort the designs on Votes, in descending order, to bring them to the top.

As for the arc plots, which I presume is what you are asking about, the tool is now available to all players at http://35.185.225.39:8888/arcplot.html. See the recent news post for more details, including the arc plot tool guide. With that in place, I'm not intending to produce them manually for all the contenders any more. But if you or someone else want to do that for future rounds and share it with players, that would be great!
Photo of whbob

whbob

  • 182 Posts
  • 56 Reply Likes
Yes, I must have back arrowed to the wrong list:( 
I was actually thinking that the Fold Change graphs above (shown after the first rapid response feedback, bead-based experiments) would continue for rounds 2 and 3.
I'm trying to understand the arc plot now.  Loops, shades and colors, oh my! Lots to take in and having fun doing it.  Thanks.
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 960 Posts
  • 302 Reply Likes
Those "fold change" graphs, which are equivalent to what we have called switch graphs in the past, will continue to be the graphical form for reporting experimental results.  The arc plots, on the other hand, are the in-silico energy model's prediction of the collection (ensemble) of all the foldings. Our lab experiments can't discern the individual foldings; they are changing too fast.  The experiments essentially measure the weighted average of all the foldings in the ensemble.
Photo of Eli Fisker

Eli Fisker

  • 2216 Posts
  • 477 Reply Likes
HOW TO CREATE A DESIGN WITH A SPIRAL ARC-PLOT

This is a recipe for how to create a design with a spirally arc-plot from scratch. There are a few discussion areas along since jandersonlee helped me along the way. I left it as is. The letter/number notation is his addition. Plus the fine arc-plots off cause. Notice that we can now make an autochain marking, so clicking one  basepairing will light up the whole chain of dependency. 


1) What design type is it?

  • If it is a same state design - place the aptamer and the reporter at a distance

  • If it is an or an exclusion design, place the aptamer and the reporter fairly close. (Not necessarily right next to each other)


In this case it was a Same state design.


2) Make both switch elements stable by putting some stabilizing base pairs around them.


Lab: https://eternagame.org/game/browse/8787272/


3) Make a drawing of all the preplaced elements


Since it is a same state design, the aptamer and reporter both needs to be bound in state 2. This gives you their complete positions with closing stems which will give you the arcs for the bottom half of your arc-plot drawing. Plus you have the neck, that you can draw for both state 1 and 2.


To make the drawing, plot in all the stem area. So the Neck, the stem closing the bottom end of the Theophylline aptamer (aptamer gate) The stem that closes the opening end of the MGA aptamer and the Stem that closes the end of the MGA aptamer.




4) Connect the roads at the roads are missing ( ? )


The roads can only be connected in 2 possible ways, if the arc plot is to give a spiral. Here is one of them.



Actually you can make the spiral even shorter, if you keep the MGA closing hairpin stem static.

(Wouldn’t that tend to leave MGA on in state 1? Isn’t it by crossing G4 to G5 that MGA turns off, or can G4 just match enough of MGA1 to turn it off?) I can try make one later, to see if I can make it work. - I managed. See the last section in the doc.  [Very nice. -j.]


[Can you label the gates/magnets/whatever - Gate1, Gate2, Gate3, ... in a spiral such that:

Theo1+Gate1 bonds to Gate2+Theo2 in state 2 in the image above

Gate2+Theo2 bonds to Gate3+MGA1 in state 1 (or to Theo1+Gate1 in state 2)

Gate3+MGA1 bonds to MGA2+Gate4 in state 2 (or to Gate2+Theo1 in state 1)

MGA2+Gate4 bonds to Gate3+MGA1 in state 2 (or to MGA1+Gate5 in state 1)

MGA1+Gate5 bonds to Gate6+MGA2 in state 2 (or to MGA2+Gate4 in state 1)


I.e. Neck1 Theo1 G1 ... G3 MGA1 G5 ... G6 MGA2 G4 ... G2 Theo2 Neck2

The alternative spiral would flip the states with the similar pairings.

...]There are a few changes in state 1. I have bolded them.


Alternative spiral


State 1 connections


  • Neck to neck  (N1 to N2)

  • MGA1 aptamer gate (G3) to MGA hairpin stem (G6)

  • Theo1 aptamer gate (G1) to MGA2 aptamer gate (G4)


State 2 connections


  • Neck to neck (N1 to N2)

  • Theo1 aptamer gate (G1) to Theo2 gate (G2)

  • MGA1 aptamer gate (G3) to MGA2 aptamer gate (G4)

  • MGA1 hairpin stem end (G5) to MGA2 hairpin stem end (G6)




5) Connect the strands in the puzzle


State 2 connections


  • Neck to neck (N1 to N2)

  • Theo1 aptamer gate (G1) to Theo2 gate (G2)

  • MGA1 aptamer gate (G3) to MGA2 aptamer gate (G4)

  • MGA1 hairpin stem end (G5) to MGA2 hairpin stem end (G6)



State 1 connections


  • Neck to neck  (N1 to N2)

  • MGA1 aptamer gate (G3) to Theo2 aptamer gate (G2)

  • MGA1 hairpin stem end (G5) to MGA2 aptamer gate (G4)





6) Pair up the above regions in the lab puzzle


First I marked the stem regions (the mmmmm) in the drawing above, in the puzzle itself. I highlighted them with ctrl+click. So Theo1 aptamer gate (G1), MGA1 aptamer gate (G3), MGA1 hairpin stem end (G5), MGA2 hairpin stem end (G6), MGA aptamer gate (G4) and Theo2 aptamer gate (G2). Then I focused only on the pairings that I needed to make for state 1.


G3 needs to match with G2 and G5 needs to match with G4. I froze the puzzle while I was working on it. Putting in the complementarities.


I made sure that Theo2 aptamer gate (G2) matched up with MGA1 aptamer gate (G3) by making them complementary. Then I filled in matching bases to them where they were placed in the puzzle to lock them in place. (Base 21-23)



I probably should have taken more of the stems themselves instead of biting in the loops. However I saw this as a chance for an easy solve.


Then I made sure MGA1 hairpin stem end (G5) matched to MGA2 aptamer gate. (G4)



Again I probably took a bit too much of the loop along. Anyway. 


I judged that I had a close to solved design and all I did was turn the mutation booster on it and let it make a solve.


Sequence: GCACCAGGUACGGAUACCAGUAAUAUCCGGGUAACGAAUGCCAGUAGGCCGACUGGAUAUUACCCUUGGCAGCCGUACCUGGUGC

Lab: https://eternagame.org/game/browse/8787272/



Here are the results:

GCACCAGGUACGGAUACCAGUAAUUGCCGGGUAACGAAUGCCAGUAGGCCGACUGGCAAUUACCCUUGGCAGCCGUACCUGGUGC,UG@25+CA@57,true


GCACCAGGUACGGAUACCAGUAAUUGCCGGGUAACGAAUGCCAGUAGGCCGACUGGGAAUUACCCUUGGCAGCCGUACCUGGUGC,UG@25+GA@57,true


GCACCAGGUACGGAUACCAGUAAUUGCCGGGUAACGAAUGCCAGUAGGCCGACUGGUAAUUACCCUUGGCAGCCGUACCUGGUGC,UG@25+UA@57,true


What I ended with:

https://eternagame.org/game/browse/8787272/?filter1=Id&filter1_arg2=8994975&filter1_arg1=899...

Sequence: GCACCAGGUACGGAUACCAGUAACUGCCGGGUAACGAAUGCCAGUAGGCCGACUGGCGGUUACCCUUGGCAGCCGUACCUGGUGC




Miniaturizing spirals by making a switching stem static instead


I have created a microspiral by making the hairpin stem closing the MGA hairpin static. I strengthened the cross over in state 1 that I left untouched. Then I turned the mutation booster (base 24-25+ 58-59) on the puzzle again and got a bunch of solves.



Sequence:

GCACCAGGUACGGAUACCAGUAAGUCAAGGGUAACGAAUGCCGAAAGGCCGACCUUGUUUUACCCUUGGCAGCCGUACCUGGUGC


https://eternagame.org/game/browse/8787272/?filter1=Id&filter1_arg2=8994994&filter1_arg1=899...


(Edited)
Photo of Zama

Zama

  • 33 Posts
  • 9 Reply Likes
Fantastic!!! I love those spirals!! THANK YOU!
Photo of Eli Fisker

Eli Fisker

  • 2216 Posts
  • 477 Reply Likes

Why don’t we have more working double aptamer designs?


I regularly get asked if double aptamer designs have shown to work. Lately I got asked if there were other examples of working double aptamer designs outside the FMN/MS2 labs. 


I only have one near working example. Design with second highest score in the Kissing loops lab Same State - Tryptophan B.


Score 86%, fold change 9.33, extra tryptophan aptamer marked

https://eternagame.org/game/solution/7656242/7691440/copyandview/


So why haven’t we had many working double aptamers since the Same State NG 2. I think for two reasons.


1) They mostly haven't done the fold exactly like PWKR's FMN/MS2 double aptamer design - having stem bases in state 1 pair up with stem bases in state 2, and pairing in the same spots.


PWKR mod, score 100%, fold change 157.06, the all time best fold change for our riboswitches

https://eternagame.org/game/browse/6369196/?filter1=Id&filter1_arg2=6415623&filter1_arg1=6415623


Notice that most of the switching bases that pairing in state 1 are also pairing in state 2. Just somewhere different. And vice versa. 


2) Too short sequence length (FMN is the smallest of the aptamers we work with, most of the other aptamers are much bigger.) Give us longer sequences and I expect that we shall spew more likely to work double aptamers.


Details on reason 1. There is a deep connection between state 1 and 2.


An arc-plot drawing based on PWKR's design as it shows in the game.

         

So just because a double aptamer design looks like PWKR's for state 2, where all the aptamers are bound, if it doesn't look similar to the PWKR design for the state 1 or some other legal racetrack pattern, it may not work.


So I think many of our double aptamers have failed for just that reason. Because the switching bases in state 1 did not connect up in the same spots as they did in PWKR’s designs. 


Now this design is very close to being a spiral design. I would love to see if a true spiral version is possible, when we get a lab that will allow us to test that. 



Photo of Eli Fisker

Eli Fisker

  • 2216 Posts
  • 477 Reply Likes

Fibonacci spirals - how perfect is that! :D


Now if anyone wonders why I’m so obsessed with spirals, here comes an explanation.


Thx to Zama for giving me the title. Also for making me look for fibonacci spirals in RNA in the first place. She asked if RNA could be fibonacci like.


The title is what she happily said to me, when she realized that RNA fibonacci spirals was exactly what she was going to get. She asked for it, I’m happy we can bring it to her thx to jandersonlee’s arc-plot tool.


The following group of designs I bring up have been outliers from what has been more normal and have caused me great wonder.



Most of this is excerpts from earlier conversations


There is one design I am particularly fascinated with. Even before the arc plots.


Exclusion - Theophylline B, fold change 29.26, score 100% and switching bases 50

https://eternagame.org/game/browse/7559902/?filter1_arg1=7612876&filter1=Id&filter1_arg2=7612876


It's an unusual exclusion design, in that it doesn't have its ms2 right next to the aptamer. It doesn't have coaxial stacking either, no static stem either. It does not follow the path of the more typical riboswitch winner of the circularized type, with a circularized arc-plot. Also it uses more bases to switch than the more usual circularized solve style (blueprint type) designs. But it is the best scoring design in its lab. 

But when I run it in the arc tool, it still get this one way road through. Although spirally.


Actually one can discuss if it is just one road or it is two separate spirals.


It kind of reminds me of a spiral galaxy (as per wiki: https://en.wikipedia.org/wiki/Spiral_galaxy)




Another way to say double spiral is double fibonacci. :)


Ok, to sum up what I'm saying: If we look at just the blueprint style designs that are circularized in solve style and arc plot, then the Same State designs can be solved using fewer bases than Exclusion designs.


This solve style is a fast way of hitting winners and make then in a structurally similar way. However there is a way to achieve a higher fold change, but also involves involves using more switching bases.


And that solve style seems to involve more gliding or sliding switches and they tend to have a more spirally arc plot.


Which reminds me of the miniaturized lab from round 101. They were made to investigate if we could delete the bases of the static stem in our bigger MS2/FMN riboswitches, without it having an effect on the design score and fold change. In one case it didn't have a negative effect on the designs, but in all the other cases it did. Some of the better designs of those small labs, showed the same stress effect, that I have seen in labs like Exclusion 1 and 3 (where the switch elements are not optimally placed in relation to each other. (FMN-FMN-MS2 or MS2-FMN-FMN instead of the better FMN-MS2-FMN)


The designs that did better in these labs that we had a hard time solving, tended to go more full moving switch - get most bases involved in switching. (Those with inverted aptamers) (edited)


jandersonlee: Eli. In arcplot you can select multiple pairs to highlight the arcs in cyan (blue-green):


GCACCAGGUACGGAUACCAGACGUUGCCAAACAGGGUGACAUGAGGAUCACCCAUGUACUGUCCCUUGGCAGCCGUACCUGGUGC


https://eternagame.org/game/browse/7559902/?filter1_arg1=7612876&filter1=Id&filter1_arg2=7612876




Eli: Wow, this is beautiful. I like it!


You even made it go through the magnet segments


jandersonlee: I want to make it so that arcplot can pick out dependency chains and possibly mark/unmark them as a group.


Eli: I love that plan


Omei: It looks to me like there are two independent dependency chains going on in this design.


Eli:


I managed to draw in the additional spiral.


It even has its own set of magnet segments too


I find it interesting that this second set of magnet segment is a flavor of the first, but different in size.


Omei ... which suggest multiple marking colors would be nice. :)


Jandersonlee: Well, what I call a dependency chain involves state1/state2 MFE pairs such as: C17=G69-C27+G56-C40 in the above


Eli: I also love that idea :D


jandersonlee: so yes, there can be multiple chains


Eli: The design of jandersonlee, the gliding switch that got the better fold change of the single aptamer designs in SS NG 2 also got a whiff of double spiral or 2 set of dependencies. (Jandersonlee’s gliding switch.)


Eli: I think this spiral concept is key to get to an even better fold change than what we can get with a single road dependency and circularized arc plots.


Another thing I find fascinating in the above arc plot is that the magnet segments for both chains gets smaller the more toward the middle of the plot they get.



One of the spirals are much longer than the other.


The smaller spiral origin alongside the bigger spiral




jandersonlee: You can find a single chain of 4-pairs for each



Eli: fascinating. It is much harder to see from the get go because not all of the 4 bases pairs up in all of the sections of the arc road.


jandersonlee: It may be enough to have one track that switches back and forth. The chain C17=G69-C27=G56-C40 has two pairs in state 1 (-) and two pairs in state 2 (=). To switch states you need to break two and make two. The inner pair must break first, so it could be C17.G69-C27.G56-C40 to C17.G69.C27.G56-C40 to C17=G69.C27.G56-C40 to C17=G69.C27.G56-C40 to C17=G69.C27.G56.C40 to C17=G69.C27=G56.C40. If you can 'roll' (or unzip/rezip) the base pair like this you don't have to break all the bases to change states. If the aptamer bases can switch first, they can perhaps free whatever blockage is preventing the reporter site from forming.


In the GCACCAGGUACGGAUACCAGACGUUGCCAAACAGGGUGACAUGAGGAUCACCCAUGUACUGUCCCUUGGCAGCCGUACCUGGUGC example, 4 pairs must break from state 1 before the first key state 2 aptamer pair can form:


Eli: "If you can 'roll' (or unzip/rezip) the base pair like this you don't have to break all the bases to change states."


jandersonlee



Eli: That’s what I think too is the potential strength of the spirally arc plot designs.


They do not fully overlap in the same manner as the more origami style designs


So one get a longer time for moving towards a switch


so also potentially greater fold change


So it gives a more gradual switch





omei: @eli how many examples of "spirally" designs can you identify?


Eli: I can mention the two late winners in the fast track lab too


Eli: Omei, I can not say for sure that spirals are going to be better than circularized designs. It is just the few outliers that have giving me a headache for long that starts show a pattern. I see the emerging of a new trend.



(Plus two from the miniature labs in round 101, those were the two designs with best fold change of all the 8 small loops labs.) 

Fold change 33.28, score 100%, switching bases - most

https://eternagame.org/game/browse/6369383/?filter1=Id&filter1_arg2=6438980&filter1_arg1=6438980



Fold change 40.44, score 90%, switching bases - most

https://eternagame.org/game/browse/6369377/?filter1=Id&filter1_arg2=6433626&filter1_arg1=6433626



Both the labs, the design structure and the arc plots are like inversions of each other.


Both designs switch over the middle of the sequence


jandersonlee: I might actually consider this a double spiral and quasi-circular combination.


Eli: Yes there is more than just one spiral


Omei: Eli, do you know if Vinnie produced similar spirals in other puzzles, and they just didn't compete well?


Eli: I don't know, but I suspect it has


It tends to use all the space


Omei: It could be instructive to identify something about the puzzles that dictated a different "best" strategy.


Eli: The arcs still almost crossover in the above spirals, although they are more skewed in relation to each other than the other arc spirals so far.


Eli: Its funny. Vienna sometimes goes in between spiral and circularized. I have found a few spirally like designs. Which isn't odd since it often do slides/glides.


Omei: If I understand correctly, it is not pattern-directed at all.


jandersonlee: I believe that it is not pattern directed.


That patterns seem to show up in winning designs is therefore all the more interesting.


Eli: It makes all kinds of arc plots


also pure circularized


So I take it too that it is random.






jandersonlee: Not circular, not spiral. partially double-spiral.


Eli: I agree with that, Jeff. It’s like the arc plot was cut up in the middle and then moved a bit.


Omei: I see a yin-yang pattern.


jandersonlee: So a third pattern of switch design?


Or just chaos?


Eli:

Quasi spiral

skewed

Doupler effect

redshift


Omei: Jeff, Why did you call yours a partial double spiral.  Just because the length of the spiral segments were different?


... and what sequence was that?


Eli: Many of the arcs in this plot do not overlap between state 1 and 2, but are exactly pushed and beside each other. So that could be a fourth strategy. Intersecting


Omei: I've been wondering whether "ideal" matching is best, with "defects" forced by the length between the half-aptamers.


That's one thing very nice about Jeff's tool -- you can change the spacing between aptamer halves, for both aptamers.


jandersonlee: I think it was https://eternagame.org/game/browse/6369377/?filter1_arg1=6433626&filter1=Id&filter1_arg2=643... "Rereedve" by ViennaUCT Score 90


Omei: Thanks.  Looks good.


jandersonlee:


Pure spiral



Double Spiral: in and out


Eli: So the two vinnie designs I shared showed a spiral can be made if one make a puzzle fold over the half and make each half of those sequence also fit with each other when they are turned the other way.


jandersonlee: Failed loop - almost but not quite looping back on itself GUAGCUAUCGGAGGAUAUUCAUACCGGAAACGGACAUGAGGAUCACCCAUGUGGCGAAAGCCUUGGGAGAAGGCUGAUAGCUAC




Eli: The above type regularly misses just a base or two (typically at the aptamer spot) so it can't fully circularize.


jandersonlee: I suspect a loop of more than 4 is resistant to switching as all bases must break to switch (edited)


Eli: Its kind of like the one of the aptamer sequences (typically also the strongest one) is split in two.


and each half of it is an overhang in one state or the other.


So it is kind of a break or weaker spot in the ring


Omei: That sounds right.  Which would imply it would be a good switch for a "memory element".


Eli: I tried make perfect matches between 4 strands. With no overhangs. And it didn't work well


This break part is something that both exclusion and same state designs share.


It is marked in exclusion, but some of the same state has it too


(Edited)
Photo of JR

JR

  • 235 Posts
  • 17 Reply Likes
I don't think spirals will switch very well, way to crunchy.
Photo of JR

JR

  • 235 Posts
  • 17 Reply Likes
  How can a spiral be crunchy. First we have to take a look at the arc plot. The top is a nt to nt map of state 1 and the bottom is a nt to nt map of state 2 connected with lines that creates an arc The olgios are marked on the rna string in red and blue. The spaces between the arcs look void but are really nt's that can't find a match. Too many voids in the spiral and no switch, the rna string looks jagged with kinks in the string, hence crunchy, difficult to match with 1 state much less switch. It's like finding your date in a really dark room and everyone is  standing in a line. You have to convince your next door neighbors to head in the direction you want to go.  The magenta arcs are the beginning and ending of the olgios and the stats below are for those pairings. The X factors at the end are a shot at the probability strength of that pair closing.
  I would think a flowing spiral or wavy arc plot may give better results, where the switch happens by stack movements and not by small groups of nts movements. 
Photo of jandersonlee

jandersonlee

  • 543 Posts
  • 120 Reply Likes
On the other hand, If you structure it right a spiral structure can enable a gradual transition that doesn't require a complete undo/redo of the RNA shape to switch.

Suppose that site B2 binds with B1 or B3 and B3 binds with B2 or B4 and B4 bonds with B3 or B5. You can get B1=B2+B3=B4+B5 or B1+B2=B3+B4=B5. to switch between the two you must unpair two sites and pair up two sites.

However because of the sliding/spiral structure it can happen one site at a time: B1=B2+B3=B4+B5 to B1=B2+B3+B4+B5 to B1=B2+B3+B4=B5 to B1+B2+B3+B4=B5 to B1+B2=B3+B4=B5.

In contrast, a looping either/or case where B1 bonds to B2 or B4 and B3 bonds to B2 or B4, you must break both sites to switch: B1=B2+B3=B4 to B1+B2+B3=B4 to B1+B2+B3+B4 to B1+B2=B3+B4 to B1=B4+B2=B3, Yes, the order switched, but think .(((...)))...(((...))). to ............(((...))). to ....................... to .......(((...)))....... to .(((...(((...)))...))). 
 
Photo of Eli Fisker

Eli Fisker

  • 2216 Posts
  • 477 Reply Likes
Excellent points, JR

"Too many voids in the spiral and no switch, the rna string looks jagged with kinks in the string, hence crunchy, difficult to match with 1 state much less switch. It's like finding your date in a really dark room and everyone is standing in a line. You have to convince your next door neighbors to head in the direction you want to go."

The more usual winning design type that is circularized in the arc plot is a slam dunk. 

"Too many voids in the spiral and no switch" 

For a same state spiral design: 

The spiral pattern is to have several strands that are in line in switching stems in the first state, swap their partner in the second state, for one further down the line than they were supposed to go - but in a fixed interval. Them all moving at once in a coreagraphed dance move, is what it may take to your next door neighbours to head in the direction you want to go. 



The spiral winners have so far been very rare. I agree that they are much harder to make work. But when they do work, I think they will be well worth the wait. 

 
Photo of Eli Fisker

Eli Fisker

  • 2216 Posts
  • 477 Reply Likes
Lab design challenge - can you invert the arc plot spirals?


I challenge you.

Can you make counterclockwise spirals for the tryptophan labs?

Can you make clockwise spirals for the theophylline labs?

If you manage, please post your images here.




Outline from discussion between Omei and I


Eli: So there is even orientation to spirals. :)

Well it is merely impossible making the opposite spiral to the one orientation the aptamer seems to like the best. :)

I'm still going to try

Omei: I can certainly see why there is a preference that depends on which half-aptamer has the strongest turnoff candidate.  But I would be surprised if it is generally impossible to use (a perhaps less favorable) one from the other half to make the spiral go the other way.

eli: Yes, it should be possible somehow, but I suspect less preferably with some aptamers.

Probably a bit of hyperbole caused by my frustration by the attempt.

omei: Do you think there is something other than which half-aptamer has the strongest turnoff sequence that comes into play?

eli: Yes, what partner aptamer/reporter there is in use. 

I have also had trouble with one specific orientation of the spinach aptamer

MS2 is stronger than FMN. I also think tryptophan is stronger than MGA

at least when it somes to sequence and binding habits

My biggest trouble with the spinach ones are that there is so little left over bases to get the job done with. Longer sequence and no problems. 

omei: Yes, the reporter is really just a second aptamer when it comes to folding.  It's only the way the experiment is done that distinguishes input from output.

eli: So I think the strongest aptamer will determine the orientation of the spiral

omei: Good point.



Photo of Poll na gColm

Poll na gColm

  • 26 Posts
  • 20 Reply Likes


It's for the FMN Exclusion A lab, but I got the spirals to pretty much go in reverse
Photo of Eli Fisker

Eli Fisker

  • 2206 Posts
  • 475 Reply Likes

Poll na gColm’s entangled aptamers causing spirals


You found a different way of making spirals. :) High five!


The spirals are made by entangling the aptamers. It actually stands out in the image highlight that I shared before. 



I dug up the puzzles behind the mirrored spiral arc plots above for the FMN Exclusion A (MGA). 


Clockwise - Big MGA aptamer side closest to FMN2 (Last FMN sequence)

https://eternagame.org/game/browse/8791824/?filter1=Id&filter1_arg2=9004919&filter1_arg1=9004919


Counterclockwise - Big MGA aptamer side closest to FMN1 (First FMN sequence)

https://eternagame.org/game/browse/8791824/?filter1=Id&filter1_arg2=9005167&filter1_arg1=9005167



Photo of Astromon

Astromon

  • 176 Posts
  • 22 Reply Likes
Very cool Pol! How did you get it to do that?
Photo of Poll na gColm

Poll na gColm

  • 26 Posts
  • 20 Reply Likes
I was playing around to see how many pairs I could make static and fortunately they formed a nice neat fold. So then I just moved one half of the MGA to the other side and built the exact same pattern in reverse
Photo of Eli Fisker

Eli Fisker

  • 2216 Posts
  • 477 Reply Likes
I can't help but wonder if aptamer selfturnoffs and spirally arc plots really are two sides of the same coin. Because they seem to come in combination. 

The new aptamers that contrary to FMN have a prebuilt in option for aptamer selfturnoff seem to spark spirally arc plots to a far greater degree than the earlier FMN/MS2 riboswitches.

(While FMN aptamers can get a selfturnoff, if a sequence that is complementary to the aptamer loop gets placed in the aptamer gate.) These FMN aptamer gate selfturnoff designs work by a similar mechanism. They typically have a bunch of pyrimidines (C&U's) in the aptamer gate that fits both in the aptamer gate, when the FMN is formed, but also is complementary to the side of the FMN loop that is furthest away in the sequence. 
Photo of Eli Fisker

Eli Fisker

  • 2216 Posts
  • 477 Reply Likes
Is orientation of arc plot spirals depending on the aptamer?


Zama spilled that she had a collection of Tryptophan A Same State Spinach spirals. She asked if anyone was interested in seeing. I was.


I decided to take a stab at guessing the spiral orientation before viewing the collection.

My bet was on that most of the spirals for that lab would be clockwise in orientation

When following the arc from outside and towards the design.

I based this on my own designs and the behaviour I seen of aptamer selfturnoffs.

I even see this preference for a specific direction in spirals for designs that do not use aptamer selfturnoffs but have a nearby sequence in the aptamer gate shut off the aptamer. I think it is the aptamer that is causing this directionality for the spirals.

Different aptamer may mean different orientated spirals.

But each aptamer may have its specific preference.


Zama's collection and all clockwise spirals

And I got my wishes. Both getting Zama’s collection and the clockwise spirals. Zama’s collection confirmed. :)

She has uploaded the collection on google docs. Here is the link.


Thoughts on aptamers and their effect on arc plot spirals

If I'm correct and the aptamerselfturnoff, or turnoff sequences are involved in sparking the spiralling arc plots, then I will expect the spirals to take a specific preference in what way they are spinning. Clockwise or counterclockwise.

Theophylline seems to spark counterclockwise spirals. But otherwise most of the spirals I have seen so far are going clockwise. (If following the arc from outside and in).

Here is an example



Preliminary prediction

- Tryptophan will spark more clockwise spirals

I base this on that all my design in the Tryptophan lab A lab, are clockwise in spiral + in Zama’s design collection.

- Theophylline will spark more counterclockwise spirals

I base the latter on Pi’s double spiral design.

https://eternagame.org/game/browse/7559902/?filter1_arg1=7612876&filter1=Id&filter1_arg2=7612876

Plus the two spiral designs I have made myself in Theophylline B Same State (MGA)

https://eternagame.org/game/browse/8787272/?filter1=Id&filter1_arg2=8994994&filter1_arg1=8994994
https://eternagame.org/game/browse/8787272/?filter1=Id&filter1_arg2=8994975&filter1_arg1=8994975


Local aptamer turnoff sequence and spirals

Now aptamerselfturnoffs isn’t solely involved in sparking spirals. In labs where the aptamer is split in two it also happens by a nearby aptamer turnoff sequence placed in the aptamer gate. (The switching stem closing the moving end of the aptamer)

For the tryptophan and theophylline aptamer which both are asymmetric aptamers, the local aptamerturnoff sequence has a habit of preferring to kick in one specific direction - all based on how the aptamer prefers to move when it gets turned off.  

Here is tryptophan kicking opposite direction between the states.



Which of the aptamer sequences are the strongest?

So maybe it is more by the movement of the aptamer itself turning off and there typically being one side of the two aptamer sequences that is stronger than the other. The strongest side will be more likely to be involved in sparking the spiral.

Theo B labs - Theo 2 is stronger - counterclockwise spiral
Tryp B labs - Tryp 1 is stronger - clockwise spiral
For FMN B labs - FMN 2 is stronger - counterclockwise spiral?


Special thx to Zama for having shared her collection. It helped clear up my wondering way faster than I could otherwise have done.