Our first multi-input puzzle!

  • 6
  • Article
  • Updated 3 years ago
I'm very excited to let you all know that, thanks to Nando, our very first multi-input puzzle is live on EteRNA!

It's an XOR gate, which senses two external inputs (microRNAs). The sequences we've chosen are actual microRNAs that have been found to be associated with heart failure. We want the design to display the MS2 hairpin if one input is present but not both, as shown here.



In the puzzle, you will see the two microRNA sequences at the front of the design sequence, as well as two fixed linker sequences, which are necessary for Johan's experiments. In states where the the microRNA sequence is ON, it will form a hairpin, serving as a mimic of the state in which an external microRNA molecule is bound. Here's the link to the puzzle! Note that Nando has already come up with a solution, so while this is one of the hardest yet, we think it's solvable by players. :)

Finally, I've generated a bunch of solutions through EteRNABot, which I've posted here for you all to look at. I would love to know if you experts spot any patterns across different solutions.

Feel free to follow up with any questions!
Photo of wuami

wuami, Researcher

  • 22 Posts
  • 6 Reply Likes

Posted 4 years ago

  • 6
Photo of Eli Fisker

Eli Fisker

  • 2217 Posts
  • 477 Reply Likes
Hi Wuami and Nando :)

Very cool!

I was playing with the puzzle and since I only got seemingly close to a solve, I decided to take a look at EternaBot's design instead.

Here I did notice some patterns. Since it’s in silico - computed, it may not turn out the same in lab. But if it does, it will help us design better.

I found that in the main part of its solves, uses the same stretch of sequence in the red stretch to pair first one microRNA, then the next. Meaning they are basically switching against the same strand. Just like two switching strand against strand in a switch puzzle. This happens despite the fact that these microRNA’s are not fully identical in sequence.

(Oligo here means microRNA)



Like Machinelves mentioned: So it probably depends on the miRNAs in question and the sequence, whether compatible for that same sequence.

I suspect so. I don't count this to work for all microRNAs. But find it interesting. I had not expected this. If the microRNAs are not too similar, we could target one microRNA to bind before the MS2 sequence and the other miRNA to bind after instead.

There are even some funny structure patterns for the No oligo and Oligo 1+ 2 states. There’s a seagull pattern for the first, and a mickey mouse with tail for one of the other.



No oligos

Has kind of a seagull structure for most of the solves.

Oligo 1

Most yellow sequences pair with the early stretch of the red sequence.

Oligo 2

Most of the green sequences pair up with either the first part of the red sequence, but some also with the last part. A few do a mix.

Most often the MS2 sequence and a stretch near it keeps identical structures in both Oligo 1 and Oligo 2 state.

Oligo 1 + 2

The red sequence mostly forms a long tail with itself. Some cases of multiloop tail or multiple stem tails.

Perspective

Now I don't know if these designs with these patterns that stand out for now will also do well in lab. However if they do well, I find this interesting. Because that means we can use both microRNAs as switching strands against the same stretch in the sequence near the MS2 sequence.

Playing with the data

Machinelves: I have never seen the data presented in this format. I would print it out and cut them into strips and then you could sort them into piles of likeness, without having to write scripts or ask a scripter to sort by xyz, but playing with in person sound just fun hehe. There is a huge variation in structure. I want this for all lab results, like the click a button for downloading a spreadsheet option. Should be automatable.

I could very much get used to this format too. :)

Thx to Machinelves for discussion and edits
Photo of rhiju

rhiju, Researcher

  • 403 Posts
  • 122 Reply Likes
eli: this is a very cool observation -- that the same 'primary' red stretch is used to bind oligo 1 *or* oligo 2 in the single-oligo conditions. There is then a 'secondary' binding site in the 'seagull'.

it suggests a strategy for general solution -- have you or others been able to go in and use this idea to arrive at novel designs?
Photo of Eli Fisker

Eli Fisker

  • 2217 Posts
  • 477 Reply Likes
By the way, Machinelves and I wrote a blog post about the Feynman challenge - which is related to the XOR puzzle. But since it got buried below a pile of older blog posts, I'm taking the freedom to highlight it here, where it has relevance. :)

http://eterna.cmu.edu/web/blog/5598341/
Photo of Eli Fisker

Eli Fisker

  • 2217 Posts
  • 477 Reply Likes
Rhiju, big thx for getting our blog post to show. :)
Photo of rhiju

rhiju, Researcher

  • 403 Posts
  • 122 Reply Likes
It was jnicol actually -- thanks John!
Photo of Eli Fisker

Eli Fisker

  • 2217 Posts
  • 477 Reply Likes
Magnet segments in multi-input puzzle

I decided to take a look at sequences in the eternabot solves of the XOR puzzles, after I had looked at the general structure patterns, and immediately recognized multiple magnet segments in the red primary binding lane sequence. And both the two microRNAs have it as their favorite binding spot. :)

Primary landing spot

There is often one G magnet or more and a general higher frequency of Gs in the red binding lane. There is a heavier presence of Gs along the whole lane. Which isn't so odd, since both the microRNAs are heavy in Cs and Us.

One landing lane, 2 planes

Like Machinelves said: It’s like a single plane landing strip with the glowing flares that say 'hey land over here!!'


Image Source

There can be one plane landing at a time, thus ensuring that when one microRNA is on, the other is not.


Image Source

Complementarity

Basically what a microRNA needs to land in a specific lane, is a certain amount of complementarity.

Not too much and not too little. Enough to form 1 or two stems. But not one real long one
since it will be harder to let go, like a really strong magnet.

Multi-input puzzle bot Strategy

This leads into something that I have been thinking about for a while, in relation to the MS2 labs, but had not gotten up yet, as I had not decided what to do with it.

So if a strategy were to be made, it should search for similarities between the two microRNAs and use the overlaps for grabbing onto to their common landing spot. Since the landing spot is decided with bases we or the robot fill in.

The destined binding lane should be using what similarities there are between the microRNAs.

If the two microRNAs used are too different, they should be given each their individual landing spot. So a jumbojet is not sharing a lane that is really meant for a model plane.

There is a big difference between if a microRNA has mostly GC pairs or it has mostly AU pairs. Actually the two microRNAs are quite different, they just happen to also have some very similar stretches. Difference in base pair content.

Machinelves: Interesting. And since the binding strip is mostly Gs, as you said that frees up the miRNAs to have a mixture of Cs & Us.

Yes, and Gs are complementary with Cs and Us, but not binding too much. The Us make sure of that.

For background read here:

Looking for complementarity

Possible third microRNA input?

Rhiju, I’m guessing what you are saying with this:

“There is then a 'secondary' binding site in the 'seagull'.”

That is potentially the landing spot that could be open for a 3rd miRNA plane landing.

I have drawn an image of how the general tendencies are for how the micro RNAs and landing spots are connected to each other. Showing that the primary lane has 3 connections and the secondary has 2.

Color coding as shown in Wuami’s document



One landing lane, 3 planes/MicroRNAs turned on

When the microRNAs are both turned on in state Oligo 1 + Oligo 2, most often both the primary and secondary landing spot end up pairing with each other, meaning that none of the microRNAs can pair there. Securing that the microRNA stays accessible, and are not hidden away by being paired up to other spots in the sequence. Basically 2 microRNAs and the second lane are competing for a spot on the first lane.

The second landing lane often has one or two C magnets and an overweight of Cs, whereas the first landing sight has G magnets and an overweight of Gs. Making this two good partners, when nothing else is supposed to be going on. So the red primary spot actually has 3 targets. :)

If a third microRNA were to be added, I think the microRNA intended for the second lane that has the least connections could be put either first or last in sequence. Because the first hairpin has a tendency for pairing with the second landing spot, which is last. So I am assuming an added microRNA first in the sequence still would have a preference to pair with the last bit of the sequence or its nearest neighbours. Similarly if it gets put last, I expect it to go for its nearest neighbours. Since I have noticed that certain parts of an RNA have a bigger likelihood of pairing with each other than others.

RNA as a shoelace

Imagine if RNA was a shoelace. The one end of a shoelace, can touch anywhere on the rest of the shoelace as long as it is loose. However a RNA design with stems are like a shoelace in a shoe, where the lace is put through holes - to hold the shoe in place on the foot. Meaning that the shoelace ends up touching itself in specific spots, but can't touch/cross itself anywhere it wants.

Similar for RNA, it often touches its own ends against each other, where as one end is more unlikely to pair with the middle of the sequence. And the longer the RNA sequence get, the more clear this get. Certain areas in the RNA have a much higher probability of meeting than others. First neighboring sequences. Then long distance sequences that are free to pair. (Thx to Drake for inspiration on the shoelace comparison - he has an awesome storyline on RNA on his profile: http://eterna.cmu.edu/web/player/34596/)

I did a drawing that includes potential future options for placement of a third microRNA. Remember, these are of course guesses.



Solving the puzzle - reverse engineering style

Rhiju, you asked if this helped me solve the puzzle. I haven’t solved the puzzle yet. Dimers I found hard to solve until I saw an example of how it was solved, and since this is a trimer... :)

This post and the seagull one, was what happened when I got stuck on the puzzle and decided to take the other way around and see if I could learn something from the bot solves. I still haven’t gotten to the puzzle solving part, as I’m having too much fun with this. :) But yes, I strongly suspect that examining prevalent patterns in bot solutions can help us with the actual solving.

I tried solving the puzzle with magnet zones alone and got it down to one small unstable spot. But as soon as I added a base anywhere, I either made the MS2 hairpins go unstable in two states or made them go stable in the two states where they were not supposed to be stable.

So I’m starting to think that it may actually be helping to have the magnets hidden a bit, for them not to react with everything across the sequence lake. To have them isolated and with a special something each, so they don’t just cross react with any fitting magnet, but the magnet inside of that sequence one wants it to find.

Magnet accelerators and sequence brakes

Imagining the feet of the bird are the strong magnet bases, like GG or CC, and the individual nucleotide choices for those strong magnet bases are potential brakes, like the wings of the bird. The particular nucleotide sequence in those strong magnet bases hold the address of where they need to go, and like the bird’s wings can also work as brakes for a more precise landing.

In terms of accelerating vs. braking the magnet’s ability to bind with other miRNAs, I'm also thinking about potential bases placed around the magnet sequence, which could be able to destine one magnet to be paired up only 2 or 3 places and not like 5 or 6, making it much more unpredictable exactly where in the puzzle a magnet sequence lands.


Image Source

The microRNA sequences

Both microRNAs carry a repetitive pattern, which I have earlier seen not do well for giving a stem the wanted structure. Meaning that if many stems in a single state puzzle carry such sequences, massive mispairing is happening.

Like many CUs or GUs on the same strand. Basically what makes them problematic is they often have extra strong bases, placed relatively close. Like CUC or GUG, and if in excess, they wish to go elsewhere. This is repeat, not of the same 1 base lined up multiple times, but rather repeat with two bases.

They are strong switch factors. Anything like GG GXG and GXXG, and the same for Cs, has a strong wish to go elsewhere, if there are enough of them. Whereas crossing Gs and Cs, like GXXC has a larger tendency to stay where they are at, compared to strands in stems that don’t have their GC pairs crossed. Again, unless there are too many of them as well.

Background article:
Strand repetition ban

Blue buffers

The two blue U zones (in the microRNA loops) do well as buffers between the microRNAs. For making loop or internal loop or hook area.

Plus the red binding lane often binds into two stems with an internal loop in between them. Those internal loops happening, are caused by the overweight of Us in the hairpin loops of the microRNAs. And I think this split of the stretches in the microRNA binding with the red lane is helping so the bind doesn’t get too stable for it to get moving on.

MS2 hairpin in multi-input puzzles

I am noticing with interest, that the MS2 magnets with lines of C and Gs, are not getting put much into use. Rather the MS2 sequence seems to be sliding in the Eterna bot solves, when it needs to get tucked away. I wonder if that will change with lab results, and if we can use an alternative solve here, for clearing up that second lane better, for a third microRNA to land on.

Thx to Machinelves for discussion, good questions and edits
Photo of Eli Fisker

Eli Fisker

  • 2217 Posts
  • 477 Reply Likes
I followed Machinelves advice of printing out the document with the images of the EternaBot solves of the XOR puzzle, and I have one more observation to add.

While a big part of the puzzle solves do follow the trend (around 3/5) I mentioned above, with the microRNA's landing in the primary binding spot, another trend is also strong.

No oligo State

- Seagull the above mentioned tendency in which both microRNA's pair up with the red sequence. (3/5 of the designs)

- Only the "green" microRNA pairs up with the red sequence. The "yellow" microRNA does not, it has more of a tendency to fold with itself, slide or pair up with the blue buffer sequence after itself. (2/5 of the designs)

Then one more tendency stands out. In half of the latter, the green microRNA pairs with the primary lane, and in the other half of these solves, the green microRNA pair with the secondary lane.

Also there is not the same tendency for both microRNA to pair up at the same spot in the red sequence.
Photo of rhiju

rhiju, Researcher

  • 403 Posts
  • 122 Reply Likes
just a heads-up -- the dev team will all be online tomorrow (Tuesday, Feb. 17) at 10-11:30am PST trying to solve this puzzle together in one room.

Hope some of you can join us to chat about what new features we might need to solve this efficiently. If you're not available, please post ideas here before or after -- we're monitoring this thread.
Photo of JR

JR

  • 235 Posts
  • 17 Reply Likes
Tagging or highlighting the red errors in the puzzles might help. Would not know
for sure until I used it though. (In that particular view.)
Photo of JR

JR

  • 235 Posts
  • 17 Reply Likes
I don't see the last status icon in PIP mode. I would like to see all on same page.
6 total but only 5 show in PIP mode.
Photo of JR

JR

  • 235 Posts
  • 17 Reply Likes
Last structural notation looks short.
Photo of Eli Fisker

Eli Fisker

  • 2217 Posts
  • 477 Reply Likes
Machinelves made some fine images that are visualizer of the connection between the states, the eternabot drawings and the truth table.



Photo of Eli Fisker

Eli Fisker

  • 2217 Posts
  • 477 Reply Likes
I tried compare the locked stretch of the sequence in the puzzle with the sequence in the Eterna bot solves, and there are several stretches that I can't make match up. The eternabot solves have another sequence for the second microRNA.

Plus the eternabot solves are only 126 bases long, where as the puzzle is 135.
Photo of Eli Fisker

Eli Fisker

  • 2217 Posts
  • 477 Reply Likes
The EternaBot ones Wuami mentions:

https://www.dropbox.com/s/ixpw2lm2clc...
Photo of Eli Fisker

Eli Fisker

  • 2217 Posts
  • 477 Reply Likes
MS2 sequence position doesn't match up in base numbers. And the bot solves seems to be shown as a mirror image. But perhaps Eterna is mirror image to how it is normally shown?
Photo of JR

JR

  • 235 Posts
  • 17 Reply Likes
Thanks. I also would be looking for the string(s) so I can upload into puzzle maker.
Photo of wuami

wuami, Researcher

  • 22 Posts
  • 6 Reply Likes
Sorry, I think I screwed up the image generation for the bot solves. It should be fixed now, although the mirror image discrepancy is still there.

@JR - Here are the sequences for just the red design portion. https://www.dropbox.com/s/dyskg6fh7q3...
Photo of Eli Fisker

Eli Fisker

  • 2217 Posts
  • 477 Reply Likes
Wuami, no problem.
Photo of JR

JR

  • 235 Posts
  • 17 Reply Likes
Starting off you may want to put the puzzle in puzzle maker mode so that
strings can be uploaded.
Photo of Eli Fisker

Eli Fisker

  • 2217 Posts
  • 477 Reply Likes
First a few suggestions for mainly old tools:

- It would be very helpful to get the paste sequence field put in. I think what will get us up to speed solving these puzzles and doing well in no time, will be to be allowed playing around with sequences of already solved puzzles. Like the EternaBot solves or each others.

- Also I like Nando's new purple highlighting and pairing strands tool. But it doesn't seem to be working here.

- High lighting bases with the black marker is rather buggy. (Ctrl + click) I can only use it some of the time.
Photo of Eli Fisker

Eli Fisker

  • 2217 Posts
  • 477 Reply Likes
Allowing us to paste in already solved puzzles, would be awesome since Wuami has already shared the sequences with us. All we need to be able to is to paste them in.
Photo of rhiju

rhiju, Researcher

  • 403 Posts
  • 122 Reply Likes
jnicol is going to deploy a 'lab' version of the puzzle on his dev server which would allow inspection of eternabot solutions. not ready for today though.
Photo of Eli Fisker

Eli Fisker

  • 2217 Posts
  • 477 Reply Likes
Awesome! I'm happy to hear. :)
Photo of Hyphema

Hyphema

  • 91 Posts
  • 25 Reply Likes
I have discovered, for me, that the highlighting tool works (ctrl/click) but only up to nucleotide 100.
Photo of Eli Fisker

Eli Fisker

  • 2217 Posts
  • 477 Reply Likes
Hyphema, big thx!

I can confirm. That explains the odd experience I had of it sometimes working and other times not. Thx for finding the why. :)
Photo of Eli Fisker

Eli Fisker

  • 2217 Posts
  • 477 Reply Likes
Another thing is that is hard to see where one microRNA begins and another ends. It is not clear as it is in Wuami's images.
Photo of rhiju

rhiju, Researcher

  • 403 Posts
  • 122 Reply Likes
note from nando: he would like to be able to paint in a structure constraint and then see the lowest energy solution & its energy given that constraint.
Photo of rhiju

rhiju, Researcher

  • 403 Posts
  • 122 Reply Likes
johan wants shift-tab to go back one states 4->3->2->1 (opposite of tab).
Photo of Eli Fisker

Eli Fisker

  • 2217 Posts
  • 477 Reply Likes
An idea that I know is for the future or a bot. But it would be awesome to specify a spot for a landing lane for two micro RNA's, then have a script search out what potential sequence will catch both microRNA's, but also be able to specify which stretch one doesn't want it pairing with.

Eg, in this setting I wouldn't want the "yellow oligo" to be complementary to anything other than the primary lane. Whereas second lane should only be complementary to the "green oligo" and the primary lane and nothing else.



What I'm coming at, is that some spots could be preset as supposed to be complementary and others not. And then some sequence could be auto generated for a starter.
Photo of rhiju

rhiju, Researcher

  • 403 Posts
  • 122 Reply Likes
looks like our playathon did not result in solutions despite having some of the best RNA design minds on the planet focusing their attention on it.

it currently takes eternabot 10s of minutes to find a solution to this problem.

we're going to see if a human prescribes a subset of residues to optimize, can eternabot find a solution faster, 10s of seconds?

if so, we could have a new style of gameplay where human give eternabot regions to optimize -- could be more powerful than humans or bots separately.*

wuami will post results here.

*Note that for more complex puzzles, eternabot fails completely -- so we may really need human+computer solvers. cyborgs if you will.
Photo of Eli Fisker

Eli Fisker

  • 2216 Posts
  • 477 Reply Likes
EternaBot, let us be cyborgs together. :)

Sincerely,
your human
Photo of Eli Fisker

Eli Fisker

  • 2217 Posts
  • 477 Reply Likes
Comments on Nando's puzzle:

http://nando.eternadev.org/sites/defa...

There is a red magnet sequence near the MS2 sequence, just like in the MS2 lab.
That's rather simple, which I actually like. I mean, it isn't at all as filled up as eterna bots. Which I think is good.

Because there are so many rather long switching elements in the EternaBot solves, which I think is not easy getting to switch. Longer stretches of switch are always harder to get moving than shorter stems.

Idea:
This leads me speculate, that it will be easier to solve the puzzle, if we have shorter stretches of single yellow bases to fill - primary lane and secondary lane in the "red sequence stretch".

JR mentioned that it has lots of red nts, not many green nts.

What JR mentions on red being more prevalent than green makes good sense. There was this same unbalance of Red to green in MS2 labs:

https://getsatisfaction.com/eternagam...

If there is an unbalance, there are fewer green and red magnets competing for each other and more likelihood that the right ones find each other.
Photo of Eli Fisker

Eli Fisker

  • 2217 Posts
  • 477 Reply Likes
Interestingly enough, Nando solves his puzzle with a red magnet segment in the primary lane and an overweight of G's there, just like Eternabot did in many of its solves, following in Nando's footsteps. :)

Similar both microRNA's lands on that primary lane. (Base stretch around 65-75)

Open for an extra microRNA...

Then there is one more thing I like a lot - like in extremely much. :)

The secondary lane doesn't pair up with the primary lane, like in many of EternaBot's solves. Instead it does something a lot more interesting.

This basically leaves second lane totally free - for a third microRNA. :)

Something is also clearly visible in Nando's image. There simply isn't anything in second lane. :)

Secondary lane highlighted:
Photo of rhiju

rhiju, Researcher

  • 403 Posts
  • 122 Reply Likes
@wuami, based on eli's question, can eternabot solve XOR puzzles that are shorter than this one?
Photo of Eli Fisker

Eli Fisker

  • 2216 Posts
  • 477 Reply Likes
And in my joy of realization, I totally forgot to say what interesting things Nando design did, that none of EternaBot's designs did. It actually used the C's lined up in the MS2 hairpin. :) Something I had wondered about why Eternabot didn't. So basically this puzzle, while dimer, is a lot in family with the MS2 labs even with its aptamers and with earlier switches too. I think it will work in a similar manner in lab.

I expect that Nando's design will do a lot better than the majority of the Eterna bot designs. Given that the lines of A's don't bring it in trouble. Its a simpler and much cleaner solve, meaning that a lot less can go wrong and mispair.

So instead of adding new green magnet segments outside of the MS2, I think that using the magnet/s that MS2 already bring to the party, will be much smarter. And it can also help secure that the MS2 is turned on when needed and turned of when not. In a much simpler fashion.
Photo of janelle

janelle

  • 29 Posts
  • 5 Reply Likes
When looking at the yellow chart at the top it shows: (using "c" for the check mark)

x x x
x c c
c x c
c c x

Now, I'm going to label the rows with Numbers

1 x x x
2 x c c
3 c x c
4 c c x

Now, when I look at the puzzles, I see in target mode reading left to right in PIP mode:

4, 2, 3, 1

The "4" puzzle being the 4th row in the chart but the 1st puzzle in PIP reading left to right for the actual puzzle.

This was really confusing for me because I thought the rows top to bottom would go left to right for the puzzle.

SUGGESTION: Change the chart to match the puzzle. This would probably be the easiest and keep the puzzle intact. I believe this will help players in reading.
Photo of wuami

wuami, Researcher

  • 22 Posts
  • 6 Reply Likes
Photo of Eli Fisker

Eli Fisker

  • 2217 Posts
  • 477 Reply Likes
I decided to make new drawings, like the ones I have made before of structure and shapes. Just this time I based it on Nando's design.

I will basically expect to see more dumbbell shapes, than seagull shapes among future working designs. Since this switch involves a stem less.



I also find the following interesting. Most of the blue and green bases are on a particular side of the stems. There was also a bias in the EternaBot solves. This is a bias that would be bad in static designs, but I expect it to do well in switches.

Photo of jnicol

jnicol, Player developer

  • 57 Posts
  • 21 Reply Likes
I updated the development server to add this puzzle as a lab.
http://jnicol.eternadev.org/web/lab/3...

Enjoy :)
Photo of wuami

wuami, Researcher

  • 22 Posts
  • 6 Reply Likes
I've tried specifying a shorter segment of sequence for eternabot to optimize (e.g. 64-73 and 92-97, like in Nando's solution). This doesn't seem to help the bot come to solution faster.

Likewise, shortening the length of the design seems to make things more difficult for eternabot.

I'm looking into introducing some of your ideas into eternabot's strategy, so keep the ideas coming!
Photo of Eli Fisker

Eli Fisker

  • 2216 Posts
  • 477 Reply Likes
Wuami, I'm wondering if removing the whole second lane, might help force Eternabot come closer to a solution like Nando's. Because all the alternative EternaBot solves will no longer be possible.

On the other hand, I think there are not many changes one can make to Nando's design and still have it work. It wants its red magnet segments. So there may be fewer legal solves, for the more simple and clean solves.

I submitted a solution that is a variation over his design, but where I pushed first red magnet segment a bit and made variations in the second. But placement of the second red one, seems to be fixed. Its those two red bases, doing the magic of forming a hairpin loop with the green base in the MS2 sequence, to break the MS2 sequence when needed. A trick that also surfaced in several of the MS2 labs.
Photo of Eli Fisker

Eli Fisker

  • 2217 Posts
  • 477 Reply Likes
Loose thought. I wonder how Nando's bot would solve the puzzle.
Photo of Eli Fisker

Eli Fisker

  • 2217 Posts
  • 477 Reply Likes
Second thought: Are there any of the usual science algorithms, that are up for the challenge?

The reason for my question is that when eterna was new, we learned a lot from watching bot designs. In particular on what not to do. But also for finding solid solves for the neckarea which then was a pain - we borrowed working ones from Nupack.

So I have a feeling that we can learn - for good and bad.
Photo of Hyphema

Hyphema

  • 91 Posts
  • 25 Reply Likes
Just a thought, in the recent MS2 labs we had discovered that the MS2 hairpin has a rather strong proclivity to form. Even when we tried to prevent it. This suggests there is a rather steep energetic "bonus" for the hairpin. When battling to get all the constraints in the XOR puzzle met I sometimes disagree with the mini map when it tells me that the MS2 won't fold. I argue that it would, and probably quite easily.
Photo of rhiju

rhiju, Researcher

  • 403 Posts
  • 122 Reply Likes
Interesting thought, hyphema.

Nando has been scheming to allow players to revise energetic parameters and share these revised sets.

Won't be deployed immediately but good to hear that one immediate application would be to one of the more important motifs in the labs...

For now, eternabot and I still think we should be able to solve this puzzle 'in silico with the existing parameters...
Photo of rhiju

rhiju, Researcher

  • 403 Posts
  • 122 Reply Likes
@Eli, NUPACK not any other of the existing design algorithms, to my knowledge, can solve puzzles with these sorts of constraints & sequence changes. wuami's EteRNAbot is the only game in town. If you do find some other tool, let's try it!
Photo of Astromon

Astromon

  • 176 Posts
  • 22 Reply Likes
Hi, is this lab still open?