Ribosome Challenge Puzzles of the Week

  • 3
  • Article
  • Updated 1 month ago
This is a conversation covering all aspects -- questions, answers, critiques, debates and musing -- of the Ribosome Challenge Puzzles of the Week
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 980 Posts
  • 308 Reply Likes
  • excited

Posted 5 months ago

  • 3
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 980 Posts
  • 308 Reply Likes
I'd like wider player feedback on the choice of marker colors that we use to indicate what choice of nucleotides can be used at each position. For starters, here are the colors currently being used.
 
              

In coming up with these specific colors, the player dev team had two primary considerations: 
  1. The colors should be distinguishable, and
  2. The colors should be suggestive, e.g. mixing yellow (A) and red (G) points makes orange (A or G).
Unfortunately for consideration 1, different people see colors differently, so some colors may be hard to distinguish for one person, while different colors are hard for another. So we can't expect one set of colors to be ideal for everyone. But we can undoubtedly make improvements. I'm pretty sure there has been research into how to choose a set of colors that are distinguishable to "most" people, so if anyone is familiar with that, please speak up.

Another thought I have is that we might treat 2-base choices (the middle line in the image above) systematically different than the 3-base combinations (the last line).  With this third consideration added to the first two, I constructed this set of colors.


           
Here I have attempted to distinguish between the 2-base and 3-base choices by making the former brighter and the latter darker.

Working in the player dev group on this, we found that it can be hard to judge any specific suggestion for an improvement without seeing it in the context of the other colors. To address that, I have created these two charts as Google "drawings" and made them readable/copyable by anyone with the URL. The URLs for these are https://docs.google.com/drawings/d/1qXcEVka68nH1sTUNc9gVnLokeQ-F4OqP5idZSvAqat8/edit and https://docs.google.com/drawings/d/1FBL1MEWYen6V_7orInubudy9tXi5Rs9SsI8NFbJCNU0/edit. So if you have some specific suggestions for one or more colors to change, make a copy of one of these, make the changes and post the new picture here. If you have used any of Google's other document types, I think you'll quickly figure out how to change a color. If not, you can always ask here.

Also, there have been numerous good ideas on how we could convey the constraints better than just colored marks.  For right now, we're concentrating on the color marks because we can do that with Eternascript, as opposed to adding new capabilities into the game. Changes to the game code are possible, but they are subject to the constraints of longer-term planning in a way that Eternascript code isn't. So keep up those thoughts, even if we can't implement them right away.

Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 980 Posts
  • 308 Reply Likes
I think this is an excellent idea! Others' thoughts?
Photo of Brourd

Brourd

  • 452 Posts
  • 82 Reply Likes
Only way to find out if it works is to create a test script  ̄\_(ツ)_/ ̄
Photo of Astromon

Astromon

  • 192 Posts
  • 26 Reply Likes
I like this great idea!
Photo of Andrew Kaechele

Andrew Kaechele

  • 85 Posts
  • 27 Reply Likes
Hi Omei, I like your proposal 1 for the change to (not A). Not complaining, I think this is all really cool, but I just spent days not being able to distinguish (not A) from (C). The only reason I caught it was by carefully running back the mutation booster changes. This is what it looks like on my screen. #26 is (not A) and #27 is a (C). I think this is a needed change.

(Edited)
Photo of whbob

whbob

  • 193 Posts
  • 58 Reply Likes
I had to change my browser to 150% zoom to begin to see the color difference between the outer circles of those bases.
Photo of JR

JR

  • 241 Posts
  • 20 Reply Likes
I would like to see what tools (if any) solvers are using to solve the puzzle. 

Photo of whbob

whbob

  • 193 Posts
  • 58 Reply Likes
Other than following the constraints color codes, I have just used the hunt and peck method which seems crude after getting a taste for the mark & mutate scripts. If there is a tool to automate mutating several bases not near one another I would like to know about it. 
Also, It would be great to know how the Arc Plot can benefit this type of puzzle. 
Photo of whbob

whbob

  • 193 Posts
  • 58 Reply Likes
As per Omei & DigitalEmbrace, I started using the glue tool ( hold down the alt key and click or click & drag your cursor over bases you wish to mark).
In addition, hold down the shift key, click & drag the cursor to mark sequential bases to mark them grey outlined.
The marking told me what already bound properly. Switching from Target to Natural was much easier that way. Resolved the two challenges a few times to confirm. Thanks :)
 
Photo of whbob

whbob

  • 193 Posts
  • 58 Reply Likes
I also found that I can eliminate all of the constraint markings over the bases. It helped me to not get distracted by the constraints for a while. I have the mark mutations script V0.7 on my booster box. When I ran it and clicked oK, the constraints disappeared. Just refresh and they will come back.
Photo of JR

JR

  • 241 Posts
  • 20 Reply Likes
I would like to see the "do not change this NT"  circle removed so I don't have to hunt down which is which. Wandering through the puzzle would go much faster.
Photo of DigitalEmbrace

DigitalEmbrace

  • 48 Posts
  • 29 Reply Likes
The development team has been working hard the last couple months creating rather ingenious constraint tools for the e. Coli rRNA puzzles. Here is a brief of explanation of each tool:

Locked bases - Some bases that are involved in tertiary bonding with other subunits and proteins are locked in order to prevent modifications that could damage ribosome functionality.

Purple regions (in the small red box upper left) - The energy models don't really know how to predict what bases pair up in these loops. Don't worry about misfolds in these areas. 

Base mutation constraints (multicolored rings) - Known species variations based on the degree of base conservation among gammaproteobacteria, which is the phylogenetic class that includes E coli. By comparing the e. Coli rRNA to its closest relatives, we have identified mutations that may be less likely to disrupt the ribosome.

One more note, be sure to click on the red box indicating misfolds in the target secondary structure so that misfolds are highlighted in red on the puzzle. Then put the puzzle into natural mode and look for stems marked in red. These are the misfolds that need to be forced apart, so try to find a mutation that a) satisfies the constraints (which you can tell by sight once you get some practice reading the colors), b) reduces the energy delta significantly (not just a few tenths) and c) causes the puzzle to significantly change its folding.

I wrote this explanation for the ribosome designs we were creating (and will be creating) but thought I would share in case it helps clarify what is going on in the weekly challenge puzzles.
Photo of Astromon

Astromon

  • 192 Posts
  • 26 Reply Likes
Photo of Gerry Smith

Gerry Smith

  • 69 Posts
  • 35 Reply Likes
In working thru Ribo Challenges, I may have come across a helpful way to reduce chances of going into a "Delta Death Valley".
Where Delta Death Valley is a design with suboptimal Delta low that you can't get out of by just adding new mutations.

When deciding which sequence to choose among a list of current low Delta sequences, I avoid sequences where there are no IUPAC mutable NT's within the red mis-folding areas.  Instead I pick a low Delta sequence with a large number of IUPAC mutable NT's.
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 980 Posts
  • 308 Reply Likes
Gerry, I just read this (again?), and it brought to mind a closely related question.

When one finds oneself in a Death Valley (regardless of whether it is a delta energy of a structural Death Valley), does this provide a good heuristic for the first step in getting out?
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 498 Reply Likes
How to see the purple areas


Astromon brought up an excellent question:

"can we get a list of all purple area bases?"

The reason why it is interesting seeing the purple areas is that they regularly hold some kcal lowering options. 

The thumbnail showing the purple areas is really tiny and most of our Ribosome challenge puzzles are large. So it has gotten rather hard seeing the purple bases. But it turns out there is a way.

LFP6: If you look at the RNA strand itself. 
They have little "halos"
Gray and purple 

Images by LFP6 With marker on


Without marker


But the purple hue don't show up with the explosion factor on. 

There still is a way to get to see the purple markers. As LFP6 said: "Temporarily unmarking them is probably the way to go at least for now"

However it takes installing a booster. Here is the one I use for this: 

Mark Mutations (v0.7) (Eli's copy)

I run the booster and say no to the option "repeatedly automark mods?"

Then the booster pulls off all the colored rings and only leave the the changed bases with black rings on. 

Now here is what the 1.7 puzzle looks like with color markers off and visible purple halo's. 




For how to install a booster see AndrewKae's intro. 

Quick Start Guide to Using Scripted Tools in Eterna




Photo of cynwulf28

cynwulf28

  • 83 Posts
  • 26 Reply Likes

Photo of cynwulf28

cynwulf28

  • 83 Posts
  • 26 Reply Likes
I am posting this here as I seem to have failed to post it to chat. These sequences are for the POTW 1.12. The original un-mutated sequence is at the top, below are the solutions listed newest to oldest. I grayed out all bases which are shared between the original puzzle sequence and the player solutions (that is, gray bases=not mutated by the player). Colored bases are player-made mutations.  These results were current as of 08-29-19. I have posted this in the hope that the pattern in player-made mutations will help in analysis. Please let me know if you have any questions.
  
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 980 Posts
  • 308 Reply Likes
I really like this, cynwulf! Do you have a script to do this? The reason I ask is that I have been thinking about doing a cluster analysis of the submissions, and it seems like this would be a great way to visualize which mutations are characteristic of the individual clusters.
Photo of Eli Fisker

Eli Fisker

  • 2239 Posts
  • 495 Reply Likes
Beautiful, Cynwulf! It stands out clear that some bases seems to be obligate changes for solving. 
Photo of cynwulf28

cynwulf28

  • 83 Posts
  • 26 Reply Likes
I've always favored this kind of analysis, but ultimately no script was used; I did all of the editing by hand (took an hour or two) in Word 2016. If I had a fast way to generate this data I'd do so regularly.
Photo of Brourd

Brourd

  • 452 Posts
  • 82 Reply Likes
You could probably use a sequence alignment software for all of that, Omei. There are better programs out there (if you want to focus more on something like sequence alignment), but I've used UGENE and BioEdit for sequence annotation and viewing. There are very simplistic alignment software associated with these, if I recall correctly.
Photo of Gerry Smith

Gerry Smith

  • 69 Posts
  • 35 Reply Likes

Photo of Gerry Smith

Gerry Smith

  • 69 Posts
  • 35 Reply Likes
Here is a variant on what Cywulf has done.  In excel, so you just need to insert sequence.  The second comparison shows pairs together.
Photo of Gerry Smith

Gerry Smith

  • 69 Posts
  • 35 Reply Likes
I'd be happy to do this for anyone - just send me list  of sequences - or share excel file.
Photo of jandersonlee

jandersonlee

  • 555 Posts
  • 130 Reply Likes
@Omei suggested that I post here about transferring a POTW solution into one of the labs. I noticed that the 1.13 POTW was a partial match to the 16Sp2 Ribosome Pilot Challenge Warm-up. By inspection it seemed that bases 11-200 in the POTW were part of the target lab puzzle, so I copied my solution from the POTW and pasted it into a text editor. Next I truncated it to 200 bases (the last base I wanted), and then deleted the first 10 bases, leaving only the sequence of bases that I wanted to transfer.  Since these were a contiguous substructure in the target puzzle, I could use the Smart Paste function in the Lightning menu of the lab tool to apply this whole sequence to the lab puzzle. (If it had not been one contiguous sequence, I would have had to do it in stages). Turning on the Soft Constraint Marker (also in the Lightning menu) I could see that the soft constraints for the lab differed slightly from those in the POTW and made one or two adjustments. The result was a sequence that matched the target structure of 16Sp2 in the area of concern. The rest of the 16Sp2 puzzle did not seem to have as many options for mods as per the soft constraints, so my conjecture is that at least some of the remaining "miss-folding" is possibly required for switching-action during the operation of the ribosome as it translates the mRNA into proteins.
Photo of jandersonlee

jandersonlee

  • 555 Posts
  • 130 Reply Likes