How Eterna players can generate 100,000 promising OpenCRISPR designs in just eight weeks

  • 3
  • Question
  • Updated 2 months ago
  • (Edited)
Well, here we are, with a new challenge and an experimental capacity for analyzing 10 times as many designs as ever before.  Wow!

I can imagine that many experienced players rolled their eyes a bit when then saw the goal of ~100,000 submissions, thinking there was no way we could get so many players to generate so many solutions.  But really, there is method in this madness.  Hear me out. :-)

There are a total of 32 puzzles, most of which have a per-player limit of 150.  That means each of us has a potential for submitting ~5000 solutions.  I know of a handful of players who can actually do this with the tools they have available.  Five players generating ~5000 designs comes to a total of ~25,000 designs, which is more than twice as many as we've ever had in the past, but it is still far short of 100,000.

How is it possible for a single player to create 5000 designs in two months?  They don't all use the same techniques, but what they all have in common is having access to programs/scripts to do a lot of the heavy lifting of generating and/or evaluating possible designs.  What I want to do is to make one or more of these automation tools accessible to 100 players, instead of only 5.

Here's an example that has been rolling around in my head the last few days.  Say you have a design you think is promising.   Maybe you created it, or maybe another player created.  In either case, you bring it up in the game and decide on a way to experiment with variations.  You mark the bases you would like to modify with the black marker, and load the Awesome booster from the booster menu.  You then choose what kind of mutations you would like to see (mutations, deletions, ...) and what screening criteria (e.g. satisfies constraints with Vienna2) you would like applied.  You click on a button and BOOM! you have a list of all the designs that satisfy your request.

At this point, you can choose to scroll through the designs to see how they look in the game and trim the list down as much as you want.  When you are satisfied, you click another button to submit the remaining designs and ... (well it won't be BOOM!; maybe you should go take a nap) your computer will do that for you.

Jandersonlee has already laid much of the groundwork for a booster like this.  But it  will take more work to turn it into something that is easy accessible and useful to any player.  And that is really the whole point of this post -- to recruit players who are willing to contribute their time and skills to making this (or other ideas) into reality.  Typically, the skill in shortest supply is Javascript experience.  But a lot more than coding goes into software development, like figuring out a really good UI, testing, organizing, cheering, etc.

The floor is now open for discussion.  Let the Force be with you.
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 1026 Posts
  • 332 Reply Likes
  • euphoric

Posted 3 years ago

  • 3
Photo of Eli Fisker

Eli Fisker

  • 2328 Posts
  • 541 Reply Likes

One more design challenge

The first two CRISPR control puzzles have a 4 way junction in their structure.

These labs are also static designs. We have already done labs with exactly this structure.

Hint: I think it will be interesting transferring some of the past working 4 way junctions.

These are the relevant past labs

The cross

Cloud Lab 15 - Triloop Buffet by jandersonlee

Triloop with multiloop core phase 1

Tetraloops and 0-0-0-0 Multiloop Phase 1

Tetraloops and 0-0-0-0 Multiloop Phase 1 Version 2

Quad Loop test

TinkerToy Retread

You can use a booster to help you transfer the sequences from the past labs to the open lab.

Script Boosters Intro

Winning designs with sequence bias

Some of these winning static 4 way designs have a strange sequence bias

With purine bases in one strand and pyrimidine bases in another strand.

This base freequency looks more like something that I would expect in a switch design.

Image taken from this background post on sequence bias in switches.

I took a look at similar designs in the lab. Interestingly a lot of its later siblings failed. So word of caution. I still want to try this out. :)

Coaxial Stacking and sequence bias - any relation?

Hypothesis list:

  • I wonder if sequence bias it will have any effect on how the reporter gets bound. 

  • I wonder if sequence bias have an effect on how effective coaxial stacking is?

Perhaps this is why pyrmidine (C & U) and purine (G & A) sequences are so prevalent in switches?

As Omei took notice, switches tend to love doing a lot of coaxial stacking with stems close to the inputs and reporters.

Anyway I intend to investigate if sequence bias will have any effect on the binding of the reporter in these CRISPR control labs.

And while a lot of sequence bias (less well mixed bases) are generally a less good strategy in static lab, for the sake of learning I’m encouraging it.

Get bias creative :)

Photo of Eli Fisker

Eli Fisker

  • 2328 Posts
  • 541 Reply Likes

Sequence bias as what I suggest, will create uneven energy distribution. Something that turns up in switches, especially the more inputs they get. However this is not good in static RNA designs. So it may be that this early very biased design is an accident. I will still investigate sequence bias. But it is likely a good idea not spending all slots on such an experiment.

Also there are a couple more relevant labs:

Eterna History Tour - Introduction to Lab - The Cross lab

Eterna History Tour - Introduction to Lab - Bulged Cross

Shape Library 103 : One Bulge Cross
Photo of Eli Fisker

Eli Fisker

  • 2328 Posts
  • 541 Reply Likes
Creating structure diversity with the Mutation-Submit booster

Astromon has been toying around with getting the Mutate-Submit booster solve close to solved lab puzzles - and with luck.

I noticed something while I was running some mutations on designs I found interesting, often even a single mutation were capable of sparking structure changes besides the mutations.

Today it struck me that we can use this to create more diversity in our solves. I was playing around with a design by Mee that I liked because it did a rather huge switch - taking use of the full space.

Structure of Mee's original design

By a single base mutation I got this:

Now it also helps if the design has a lot of weak basepairs like this one, as this will spark even more structure changes.

So here it goes:

1)  Pick one design that interest you - One of your own or someones elses.

2)  Mark all the bases that are not the MS2, locked bases or static stems.

3)  Run the Mutation-Submit booster

4)  Click PRUNE

5)  Go through the results and see if something interesting has happened.

Get inspired!


Photo of Eli Fisker

Eli Fisker

  • 2328 Posts
  • 541 Reply Likes

Fractal Symmetry, Butterflies and Switches

Here comes a longer storyline as I want to share the path that lead me to this conclusion. It involves butterflies...

  • Switches don’t just like regular symmetry - they like fractal symmetry.

  • Coaxial stacking causes symmetry. 


I suspect we may be able to use this to turn off CRISPR

Base repeats

I have been playing around with CU and GA repeat sequence that I have seen turn up very frequently in switch elements and designs. To some degree also CA and GU - while I think the former are the best. But perhaps even switches need a little variation despite they seem so fond of repeat sequence. :)

I have been playing with 3-ways and 4-ways as a means to get my favorite switch repeat base sequence into lab solves.

I have also been playing with complementary sequences that repeat but have less base bias.

I was trying to double the blueprint of jandersonlee’s gliding switch. But there were not enough space for it with two MS2’s. I ended making as big as one as I could without the extra MS2 that I wanted in.

Original - CAUG repeats - Single aptamer design type with best fold change

If you stick in an alternative MS2, then these repeats should get updated to match the bases at the marked position in the MS2.

Since there wasn’t space for me to double the blueprint of the above design, I did the second best. I added an extra switching hairpin. I started to play around with one MS2 hairpin and 2 switching hairpin stems + a neck of identical sequence as the bottom of the MS2. In other words, two times two switching hairpins. Those will have a lot of combo options. They can match up all 4. One could also make two different sets of switching stems. (Illustration of the later below)

001 2jpg

I played with the 4 switching stems of the same kind interacting with each other in both Exclusion and Same State labs. Here is one with extreme sequence bias. I tampered with the alternative MS2 too, to make it even more extreme. This is pure CU and GA repeats.

I drew up a few magnet blueprints. The top one is based on the design above. Notice this solve style allowed me to move the MS2, from being right after the aptamer.

004 2jpg

The bottom one is based on the design below.

It kind of ending up as a switching cross. Just folded inward.

I did a similar experiment in Same State designs.

003 2jpg

I realized that it didn’t really matter at which of the 3 hairpin positions I put MS2 at. I could solve it nonetheless. What I wonder is what lab will think. Will it prefer the usual middle one as I expect or will it be possible solving it with the one of the side MS2’s?

Double coaxial stacking

Omei’s talk about coaxial stacking in ON state, made me realize that I will probably not gotten enough bang for my buck in those of my experiments where I placed two coaxial stacks next either MS2 or the aptamer gate. Because adjacent stems only coaxial stack two and two. 3 stems make no double coaxial stacking pair.

Even more. It is not just even length stems and repeat sequence that are the cause of symmetry in switches. Coaxial stacking is causing it too. As coaxial stems are symmetric by nature.

So I decided to take one of my experiment designs, and split the static stem in two static stems and move the one part so both of them would be doing coaxial to a stem in the ON state. One at the aptamer gate and one next to the MS2.

Double coaxial stacking potential


Structurally this looks remarkable like Brourd/Nando’s TB OFF switch solve. Except that this is an ON switch and the MS2 is not split between two stems.

I decided I would throw in complementary switching stems for the coaxial stems, just as in jandersonlee’s gliding switch.

Here is the result. Now the stems are no longer static. The whole switch bubble is changing structure between states.

It is doing a similar movement as I have long liked riboswitches to do. Albeit this one is a bit skewed.

Have state 1 fold perpendicular to state 2

Generally Past example from good riboswitch lab. Good switches generally like to fold with symmetry to them.

This was also why I could fit this switch onto a fortune teller game.

Image borrowed from Blueprint of FMN/MS2 riboswitch

CRISPR Butterfly and the OFF switch

Now the CRISP part of the puzzle itself carries almost a butterfly pattern. Albeit an open ended one. I find this immensely interesting.

Which opens up for an old story. Boy and girl want to meet. Things happens. In this case preferably an OFF switch button for CRISPR. :)

I believe a fold over the middle (horizontally) will do wonders for a DNA guide kickoff.

1, 2 and 3 order fractal level

For each level the number of switching stems double

Middle folding line for turnoff

Puzzle demonstration of second and forth order level

Cynwulf has made a big puzzle that does the movement I kind of imagine for the double Butterfly image. With two halfs of the puzzle meeting in the other state.

Plus while looking for the puzzle I realized he has already made proof of concept for level 4 for this butterfly pattern. I’m aware this is simulation and this is not lab. But this is impressive.

Notice that the motif itself is coaxial to the next motive.

While the puzzle switch, there isn’t much structure change. There are only 14 switching stems in the main middle part of the puzzle and not the 16 I would have expected. This is due to the motifs sharing coaxial stackings.

Photo of Gerry Smith

Gerry Smith

  • 92 Posts
  • 50 Reply Likes
Eli -

I think I have found far more biological advanced life form patterns..and both within the same design:



Here is my analysis:

E is first letter because it is the first letter of the largest animal.

T is the second letter because T comes at the end of both of these animals.  So 2 t’s mean 2nd letter.

E is the third letter because it is the third letter of the largest animal (do you see the alternating symmetry pattern?!?)

R is the fourth letter because it is the first letter of second name of the smaller animal.

N is the second to last letter because because it is the second to last letter in the largest animal AND the are two “N’s” in bunny!  Confirmation of second position.

A is the last letter because A is the only letter left that appears in both animal’s names.  Also within the largest animal’s name “A” starts the word “ant”....and we know that ants come after elephants and bunnies.


Of course the pièce de résistance confirming this analysis is the remaining un-used letters from both animals names.

We have remaining, the letters:  B, U, Y, B, B, I, L, P, and H.

When rearranged this spells “HUBBY BLIP”.

Which is exactly what my wife calls time on eterna, as in he is having a “hubby blip”.


So can there be ANY DOUBT that this is the design that turns CRISPR on and off?
Photo of Eli Fisker

Eli Fisker

  • 2328 Posts
  • 541 Reply Likes
Gerry - big LOL!

I love your lab design analysis. It's quirky, cute and hillarious.

I can see you had good fun. I had good fun just reading it.

Alternating symmetry pattern - ETE?

By the way, I like the design. It is doing an inversion between states.

Left side - T standing on its head.
Right side - T standing up.
Photo of Eli Fisker

Eli Fisker

  • 2328 Posts
  • 541 Reply Likes

Ideas for how to kick off the DNA Guide

I’m considering potential strategies to turnoff CRISPR by kicking off the guide DNA.

I know we aren’t supposed to actually try turnoff CRISPR this round. But instead of waiting to see what may cause that action by accident, I thought we might as well take a stab at doing it at purpose.

So here comes my thoughts. I will be interested in hearing what ideas you have.

What in the CRISPR part of the puzzle looks like an OFF button?

There are 3 magnet sections that looks very much something that could be involved in a switch. Now this is far from the Guide DNA.


These are sequences that we can easily find something to target with. They resemble what is in both FMN (GAGG can be made by adding an extra base) and MS2 (GAGG CACC) Even should there not be direct complementarity between a GAGG and a CACC, it can be achieved by a middle sequence and word change gaming. CUCC or GUGG.  

Make sure that the RNA Guide can’t do coaxial stacking with a neighbour stem

Breaking the coaxial stacking next to the RNA (Kind of the CRISPR input that binds to the guide DNA. This RNA is doing a coaxial stacking with the stem beside it.

Inputs love to be on when they are next to a stem they can do coaxial stacking with. So to kick off the GUIDE DNA, one should break the coaxial partner stem.

Here there are single bases between the RNA binding to the guide DNA. The RNA "input" can not get the coaxial stacking bonus. So I guess it will be less likely to bind tight to the guide DNA.

Target the guide RNA with something complementary

A different approach could be to use the FMN NG puzzles that leaves unlocked bases before the aptamer. One strand could be made complementary to the RNA pairing with the guide DNA.

Complementarity between top and bottom butterfly

Since the bottom butterfly is locked, that leaves changing the top butterfly. This may not encourage the most switch like sequences.

Make the middle part of the design unstable

Introduce weakness in the middle of the design (vertical). Related to above suggestion also.

This will increase the chance of the two half of the puzzle wanting to meet. And big action happens.

A crossed GU will do wonders to break up the bottom and top half of the puzzle and help parts of the opposite ends meet.

If you can make the GU’s cross in the other state too, I think double bonus.
Photo of Eli Fisker

Eli Fisker

  • 2328 Posts
  • 541 Reply Likes
Fractals and switches

Fraktals I in particular like in relation to switch RNA are the Minkowski fractal plus the bow tie fractal.

The latter resembles the butterfly structure itself in overall shape, where the former resembles the individual stems more.

I like this particular image a lot. I think that would be kind of what a level 4 fractal switch would look like, if it wasn't sharing all those coaxial stacks as in Cynwulf's puzzle.

Thinking about it, this image actually also looks like a butterfly. Just like a bow tie fractal and a Minkowski fractal too.

Butterfly fractal! :)
Photo of worseize


  • 29 Posts
  • 12 Reply Likes
Photo of worseize


  • 29 Posts
  • 12 Reply Likes
2 lines (states) need to be horizontal or as follow in picture decreasing angle 
It is better if first horizontal line starts from begining of plot and angle should be less than 0 degree or 0 . 
Photo of worseize


  • 29 Posts
  • 12 Reply Likes
100 eterna Score in different labs designs with their melting plot 
Photo of Eli Fisker

Eli Fisker

  • 2328 Posts
  • 541 Reply Likes
Chemical modifications as a key to the lock, in a CRISPR switch?

Lately I have been fascinated chemical modifications of RNA. As akin to epigenetics in DNA.

I have been wondering. What if a switch could be inactivated by a chemical modification? As a way to turn off a CRISPR switch. I asked Rhiju. 

rhiju das  check this out 

6-methyladenosine-dependent RNA structural switches regulate RNA-protein interactions

eli  Wow, beautiful. My mind is blown. :)

From what I understand, it is the chemical modification in itself that is considered a switch. 

That is controling weather or not the RNA binding proteins can gain access to bind to the RNA binding site in the mRNA.

So I have heard about molecular switches. (Super small) Then there are our eterna RNA switches, that changes structure on binding with a flourescent molecule. (Big).

So this chemical modification switch is kind of inbetween. The chemical modification itself is small. But it causes a structural change in the RNA, which changes how accessible this RNA is for interaction. So this is kind of inbetween in size.

What I was kind of imagining was a chemical modification as a lock on some chrispr RNA medicine switch. As a key to unlock activity of the switch. What this paper is doing is showing that nature is already itself using chemical modifications for causing RNA switches. Which changes the activity/accesibility of the mRNA.So what I wonder is if these two concepts have been combined? CRISPR switches with chemical modification locks and keys.

Rhiju reacted with a grin. 

I'm also thinking that one may not want to use the most overall used chemical modification system in the body as if things go haywire, one will not want to affect the whole organism. 

rhiju das: correct: modification sensors don’t exist but could be potentially quite powerful at understanding RNA biology and perhaps at some point in treatments.
*treatments based on ‘smart’ RNA medicines

Anyway, the paper Rhiju shared proves the concept that a chemical modification can lock and unlock an RNA structure. So I can't see why a chemical modifications can't get stuck on a CRISPR rna switch too. As an extra level of a safety lock on the CRISPR scissor.