Eterna 4: What are your BIG ideas? (Dev chat followup)

  • 3
  • Idea
  • Updated 12 months ago
  • (Edited)
Following up on conversation in today's dev chat: http://eternawiki.org/wiki/index.php5/2017.12.15_Dev_Chat

Eterna is at a really exciting moment - we have the potential to gain the resources and opportunity to completely rebuild Eterna - providing more robust infrastructure, more platforms, better global availability, and the prospect of making significant changes to the user interface, new features, and so on. The dev team wants to know the biggest dreams community members have as preparations are made for such groundbreaking changes. What have you always wanted to see from the Eterna website and game? What would make Eterna more fun? What would allow you to be more effective? What could help facilitate puzzle solving, scripting, testing hypothesis, or collaboration? What did I not even mention? Anything's fair game!

List your best, biggest, visionary ideas for the next generation of Eterna here!
Photo of LFP6

LFP6, Player Developer

  • 613 Posts
  • 109 Reply Likes
  • Excited

Posted 2 years ago

  • 3
Photo of rhiju

rhiju, Researcher

  • 403 Posts
  • 123 Reply Likes
There's a cool discussion thread started over at https://getsatisfaction.com/eternagame/topics/opentb-results on pattern languages.

as has been pointed out, one possibility for the future of eterna is to change its interface to involve manipulation of colored beads that represent these differnt domains. There could be an arbitrary number of colors, not just 4 nucleotides, to cover A, B, C, D, E,  ... R, ... and some kind of change in symbol to represent A' [or rc(A) in jandersonlee's notation]

 I'd like to point out that we could even provide an approximate folding engine that simulates the structure of such a pattern given some coarse-grained "energies" for the bead attractions (which might even be tunable by the player). And could even simulate what happens when a small molecule binds, given an appropriate energetic rule set.   And then designs of those patterns could be compiled into actual RNA sequences by eternascript.

I suspect that such a bead interface would even let us design 8-bit adders... its always been a dream of ours to solve that 'feynman challenge', and i recently did a back-of-the-envelope calculation suggesting that we could make such a device with ~2000 nucleotides, and there might even be experimental ways to synthesize lots of such designs for testing. 
Photo of Gerry Smith

Gerry Smith

  • 65 Posts
  • 31 Reply Likes

A chatbot assistant as an intelligent front end for database inquiries such as:

Within the TB Round 2 A*B/C^2 Dec and Inc, show me the histograms and designs for all pairs:

     ·   with only 1 nt difference

     ·   that nt change being an A versus a G

     ·   with that nt difference located in the reporter complement

                  OR

Show me the designs for the top 10 pairs with 3 nt changes with the highest Global Fold improvement and show me what and where those changes were.


Database source would be google files that contain all that data and links to both designs and histograms (see below sample of this type of file). 

A*B/C^2 Dec

https://docs.google.com/spreadsheets/d/1ndNdfLsNXk0QgHBHdTBhB2g80OpKWrO5Atf3_UnDU3M/edit?usp=sharing

System would need visual pattern training to use the links to Designs to recognize forms such as “reporter complements”.

Photo of Eli Fisker

Eli Fisker

  • 2223 Posts
  • 485 Reply Likes
Input stamper as start for playing around with element sections

@Rhiju, you said the following:

"There could be an arbitrary number of colors, not just 4 nucleotides, to cover A, B, C, D, E,  ... R, ... and some kind of change in symbol to represent A' [or rc(A) in jandersonlee's notation]"

A start of playing around placing in full elements could be to give us an input stamper. Not unlike MS2 stamper - but instead stamping the input complements directly in the puzzle.

  • It could be stamped in full length or limited in size by placing two black highlight rings it could land between.
  • Or one could paint with the gluetool the landing position.
  • When it was landed, it could then be cycled through mutations so one get the parts of it one wants.
Then we can get to the fun designing part faster.
Photo of Atanas Atanasov

Atanas Atanasov

  • 42 Posts
  • 15 Reply Likes
I would like more marker tools.
We currently have ctrl+click - a black highlight of the bases.
Having more than one type of marker would help with tracking multiple binding sites etc.
A simple option is to have more colors in addition to the black.
A better option might be instead of highlighting the base circle, to have lines following along the shape of the structure with some color options (similar to what Eli draws in the screenshots).
(Edited)
Photo of Eli Fisker

Eli Fisker

  • 2223 Posts
  • 485 Reply Likes
I like your idea with highlight.

I recall Cynwulf wishing for a way to highlight and seperate switch domains.

Today I realized there is an additional way we can use highlight, to seperate switching areas from each other. Just painting with bases.

Photo of whbob

whbob

  • 190 Posts
  • 57 Reply Likes
There is also the shift-click & drag mouse function. It is for a string of bases and it puts a gray shade along the selected string of bases.  It can be used in separate places of a design sequence by using it in different states.
Photo of Eli Fisker

Eli Fisker

  • 2223 Posts
  • 485 Reply Likes
Additional functionality of existing highlighter

I would like the black ring marker to function the same way that the base lock does. One can drag the mouse over a sequence of bases and they will be locked or unlocked. With the control click for highlight one has to do it one base at a time. This will save a ton of time.
Photo of LFP6

LFP6, Player Developer

  • 613 Posts
  • 109 Reply Likes
I've had a number of ideas that I've either thrown around or found interesting based on discussions with other users over my time as an Eterna player, so I'll try to compile them here (not going to bother searching for the original sources unless prompted, as there's a lot of them and I think these descriptions should suffice).
  1. (Obviously) Rewrite everything with modern web technologies
  2. (Obviously) Mobile-friendly website and game
  3. Internationalization (already being discussed)
  4. Server-side energy model to speed up long calculations (if/assuming we can take advantage of high performance computing and/or distributed computing)
  5. GPU accelerated energy model (this would probably require rewriting or substantially altering the energy models, and I don't know enough to say if it would actually bring speed increases, but if it would help it has the potential to carry significant performance increases)
  6. A custom energy model based on Eterna data is something that has been discussed (I believe there is some work being done in this vein)
  7. Game mode that allows players to work on smaller subsets, more modular, or more generic problems (such as Rhiju mentioned). This would allow better mobile UIs (less stuff to fit on the screen), better onboarding/learning curve/initial involvement for new users (handling large complex problems can be intimidating, unappealing, or just too difficult for new users), as well as allowing for a method of gameplay when a user doesn't want to spend as much time or wants to see results faster. I'd also be interested to see if there are other types of "real science" tasks that could be added to serve this purpose (of course we do want to be careful about lack of focus/scope creep having a ton of different game modes leading to Eterna being less effective and harder to maintain).
  8. Tiered progression, so that instead of requiring a player to complete all tutorials to access all labs, as various set of puzzles are completed, "sets" of labs (or other "real since" problems a la #5) of increasing complexity become available.
  9. Simultaneous puzzle play between multiple users for teaching and hashing out ideas.
  10. Improve gamification mechanisms (notably points). Currently, the bulk of points are awarded for challenge puzzles (especially as lab coin is broken, but we're working on that!) which currently have minimal scientific or training value, areas such as analysis and scripting aren't directly rewarded, and there have been complaints from some users in the past that Eterna has become too much science and too little game (namely in the labs). I think it would be beneficial to rework how points are awarded so that they more directly reflect the types of activities that are beneficial for research in Eterna as well as the user's engagement and experience in Eterna as a whole. Other aspects of gamification should also be considered too, as points alone don't constitute a good game, but I don't have any specific suggestions off hand.
  11. Revamp the scripting interface. First off would be the need to improve it security-wise. Past that, it would be great to have more editor features (simply updating/enabling more features in CodeMirror would help), running the code through Babel to improve browser support for new JS features, the ability to split code into multiple files, some form of revision control/versioning or even Git integration, maintaining information about the original script when making a copy of it, better filtering (ie, search by user, filter by script type, more detailed script types/tags, etc), and so forth.
  12. A wiki revamp is something that's been discussed a few times in a few different manners and contexts. Since the wiki isn't used much, I think we'd really need to consider what content the wiki currently contains or should contain in the future, and what kind of functionality we use or want out of it. This way, we can restructure the wiki to account for this or find/build another tool entirely that would be better suited. The way I see it, its place is/would be to hold and organize definitions, techniques, instructions, or other reference material, as well as puzzles, labs, or other projects. This leads me to my next point...
  13. Better organization of information. This is also something that has come up in discussion a few times. We have all sorts of guides, analysis, utilities, and so forth, split between pages of forum posts, Google docs, wiki pages, and so forth. It would be a great help to have these available in a format that is more accessible, digestible, and generally easier to find what you're looking for. On first glance, I'm tempted to say that the answer might just be better utilization of and updates to the wiki, as well as a better forum setup. Will this do everything we want? Don't know, and is something that should be discussed.
  14. Change forum platform. GetSat is a great tool for a use case that is a bit different from ours. I'd advocate setting up a Discourse forum and porting relevant content to there.
  15. Change issue tracking platform. Per #13. I'd probably suggest JIRA, which we should be able to get for free once we go fully open source. Other options would be The Bug Genie or OpenProject
  16. Overhaul groups and collaboration. Currently groups are basically glorified comment streams. I'd love to see these reworked to become "working groups", where users can come together fore some specific purpose and to have the tools to be able to collaborate in working on whatever project(s) they have in mind. Features could include task lists, wiki/forum/documents, group-created/submitted puzzles/labs/solutions/scripts, announcements/blog, and so forth. It can also find some of its structure from social networks...
  17. Integrate professional social network features. There are some neat platforms out there like HumHub and eXo Platform that we can learn from. Yes, we have a newsfeed, it was even on the front page at one point, but there's a little bit more to consider. One limited factor is that we should not only see things like comments, but a more holistic view of what's going on that affects the user, activities performed by users or on content of interest, even groups like #15 (in this way, groups are able to act as a hub/subsection/filter of general site activity between/localized to users connected around a purpose), and having easy methods of viewing the current state of content the user is interested in...
  18. Better dashboard/portal. When a user logs in to the site, they should see information about what's been going on, the status of items being tracked, as well as acting as a launching point for new actions. There's been a start on making this happen with the current homepage, but I think more could be done. There could be more included on available labs and their statuses, in-progress solutions, ongoing projects/banner-like news, group project tasks/statuses (#15), statistics from recent labs, links to recently viewed or worked on submissions/scripts/documents, and so forth.
Photo of LFP6

LFP6, Player Developer

  • 613 Posts
  • 109 Reply Likes
19. Make the chat able to act more like Slack
Photo of Gerry Smith

Gerry Smith

  • 65 Posts
  • 31 Reply Likes
In switch puzzles where target shapes are known, to assess which AU orientation to use for any pair, it would be helpful to see the total 3 state Natural/Target energy difference for the whole thread using AU's to fill in the connected pairs in other states.  So we could just flip AU orientations back and forth and see the consequential difference for the whole thread for all 3 states at once.

Probably good switch AU probability rule for algo's too.
(Edited)
Photo of Astromon

Astromon

  • 184 Posts
  • 23 Reply Likes
I like all these ideas! I would like a script that will find all possible kernal atractors, for a TB-olego and the ms2 for instance. Thanks!
Photo of whbob

whbob

  • 190 Posts
  • 57 Reply Likes
Improving the detection and ratio's of multiple oligo's:

We are beginning to experiment with new small molecule dye targets.
Improved illumination intensity has been mentioned as a desirable feature.

With regards to the OpenTB labs, I am wondering if selective dye targets would work.
What if TB-A and B could be dyed Blue and TB-C could be dyed Yellow?

The combined design sequence should produce a hue of Green.
A bluish green would indicate a higher concentration of TB-A and B.
A yellow green would indicate a higher concentration of TB-C.

A discriminator can recorded the measurement in angstroms.
Resolution would depend upon frequency and not intensity.

I think that this would be a very interesting experiment.   
Photo of Poll na gColm

Poll na gColm

  • 26 Posts
  • 20 Reply Likes
During labs I like to submit solutions that satisfy all 3 engines; vienna, v2, and nupack. Would it be possible to have something that looks at all 3 engines at once, and gives limited feedback as to whether all conditions are satisfied? Because at the moment I have to manually flip between all 3 engines and that resets the "undo" function, so sometimes I lose track of what I changed and lose progress.
Or alternatively introduce something that doesn't reset the "undo" when I change folding engine?
Photo of Eli Fisker

Eli Fisker

  • 2223 Posts
  • 485 Reply Likes
Display alternative MS2's as stable ingame

We are doing round 2 of Single-input switches, revisited

The design scoring best in the first round (max fold change error 1.25) is a design using an alternative MS2.

The MS2 is showing up as not present and not turned off (red boxes) as the game is only made to recognize the original MS2, while the puzzle is conforming to structural constraints (green boxes). 



I would like to see the alternative MS2's showing up as satisfied too. 

I think this will greatly improve our amount of winners, because more players will start using the alternative MS2. Some of the new aptamers will work better with an alternative MS2. 

Further benefit would be that we could use the mutation booster to brew up more variations. If the puzzle is recognized in game as stable, it will allow us to do a prune and be left with the stable mutants. 
(Edited)
Photo of cynwulf28

cynwulf28

  • 80 Posts
  • 22 Reply Likes

Hello all, this idea could easily warrant its own post, but I felt that it belonged here.

  I commented to Eli Fisker earlier that, “the hard effort of the Lab techs and Devs isn't always obvious on the site...and so sometimes it seems like they aren't even there”.

   This wasn’t a criticism of their efforts, having done wet-Lab work with RNA myself I am keenly aware of the time and effort required. I rather feel that these efforts don’t always transduce to the surface of Eterna in a way that is clearly noticeable like players.

      The Devs and those members at Das Labs excel at getting the word out about new projects, when the need to recruit player help is at its highest. Similarly, the Eterna Teams excel at providing the conclusions of their research to players...information which is essential to improving player performance.

  However, in between these termini there is the long and often arduous process of experimentation, the most essential part of any lab. It is during this interim that player interest can wain in a particular project and even Eterna in general (I suspect this to be one reason that some players have focused on player-made puzzles rather than the labs, the gratification gained through puzzle solving is both more readily obtainable and it is persistent). In short Eterna has a problem with mid-lab slump.

     I believe that multiple solutions are ultimately needed to fix this, but I will propose one here. Anyone who has ever watched one of Space X’s Launch Livestreams knows how engaging they can be. The speakers maintain a near-constant engagement with the audience and have a contagious energy. And there is one more thing they have...clearly demarcated AND celebrated goals/milestones. Check out this video from the falcon heavy launch and pay attention to that timeline at the bottom of the screen   

                                             https://www.youtube.com/watch?v=wbSwFU6tY1c  

   Each Lab has milestones and a tentative timeline. I think that if those milestones were represented more clearly on the Eterna homepage and if each milestone were celebrated when they were met, then players would be able to better see that there IS action happening, that progress IS being made. I’d be glad to hear other suggestions/thoughts from the rest of you, and to those members of the Eterna development and Lab teams Thank you for all of the hard work.
Photo of Andrew Kaechele

Andrew Kaechele

  • 85 Posts
  • 26 Reply Likes
Maybe a scrolling banner on the homepage with:

1.) Current points leader.
2.) Most time as points leader.
3.) Other past points leaders.
4.) Most points in the last 6 months.
5.) Most points in the last 3 months.
6.) Most points in the last 30 days.