Reaction Design - ideas to reduce confusion

Case number:845813-2009896
Topic:Game: Tools
Opened by:Susume
Opened on:Friday, June 26, 2020 - 19:07
Last modified:Friday, July 10, 2020 - 22:21

Some suggestions to make Reaction Design tool less confusing (not all my ideas, some swiped shamelessly from vet chat):

1. When I choose a new reagent, do not reposition the base element in my puzzle - leave the base where it is, and just substitute the reagent. Show the "candidate ghost" aligned on my existing base. If I am changing bases, try to align so reagent 1 will at least be near where my current reagent 1 is, etc.

2. When showing the base and reagents in the mini-windows in the tool, start all of them at the same angle: e.g. choose a default position for the base, and show the reagents in the orientation they would be if attached to that base in that position. Then we can see what is different between them. ["Oh these have the same 2 hexagons, but one's blue elbow is up and the other's blue elbow is down."] Even better would be to show them in the orientation they would be if we added them to our current base at the current camera angle. If we can't see the patterns because things are all turned differently, you are not getting our best pattern-recognition skills.

3. Nice to have: a way to reset the reagents to default display position if we have spun them around in their mini-windows, by re-opening the tool or with a "recenter" button or something. If their initial position is based on the current position of our base in our puzzle window, and we move our puzzle camera, being able to align them all to the new camera angle would be great.

4. In the current puzzle, column 3 is just chirality of the base. It would be less confusing to have two bases that differ only in chirality in the upper window, and not have column 3. Save columns for things I am actually attaching to my base. 2 columns = 2 attachments, etc. If you want to offer us 3 bases, each in 2 chiralities, just list them as 6 bases in the upper window. If they are all aligned the same way in the display, we'll figure it out.

5. It would help a lot if the base and each reagent could have its own color of highlighting in the "candidate ghost". Then I know reagent 1 is the purple bit and reagent 2 is the cyan bit, or whatever. Maybe you could give us RGB pickers (in a secondary dialog) for each one, or put their glow colors in the options file or something, since different people find different colors hard to see.

6. In the tool, show us which one in each set is currently in the puzzle.

7. In the tool, give the bases and reagents numbers so we can remember and talk about what we have tried. ["I'm using base 2 and reagents 1 and 3, what are you using?"] If we can't talk about what we're doing, you are not getting our best collaboration skills.

(Fri, 06/26/2020 - 19:07  |  9 comments)

nspc's picture
User offline. Last seen 10 hours 20 min ago. Offline
Joined: 03/26/2020
Groups: Go Science

Instead of current menu, maybe we can have a contextual menu in each molecule part to change it, like the menu to choose a sidechain in proteins.

Joined: 06/20/2019
Groups: Go Science

Yes, I do agree that we should have something like nspc said.

I would also like to add on that I think the supposed hypothetical contextual menu should be so that only every oxygen (to accommodate for ether groups), carbon, & nitrogen (to accommodate for tertiary amine groups/quaternary ammonium cation groups [e.g. acetylcholine - an important physiological ligand in the nervous system containing both quaternary ammonium cation & ether functional groups]) atom has its own set of viable functional group options. This means that when designing ligands, I think it should be so that every carbon, oxygen, and nitrogen atom is its own molecule part.

I think that how this hypothetical contextual menu would supposedly work is so that, since each and everyone of those atoms can make their own number of bonds, so there should be a list of available options for each bond.

I think that this also means that there should also be a way to select exactly which bond on that atom to be modified, so each bond of the atom should be indexed, and respectively and --distinctively, including but not limited to towards color-blind people--, highlighted in the 3D rendering of the atom itself in the ligand.

Because of the large number of functional groups and chemicals and known organic reactions (e.g. hydroxylation, haloalkanes, hydroxylamine, Rosenmund-von Braun synthesis, Kolbe nitrile synthesis) in organic chemistry, this will open up a larger number of possibilities and combinations for Foldit players to recognize and memorize and compare.

For example, I think Foldit players should be able to replace white hydrogen atoms with purple atoms.
This is called hydroxylation, which is replacing a hydrogen atom in a C-H bond with a O-H (hydroxyl group, the top of serine), physiologically usually via iron compounds as a catalyst in the enzyme cytochrome P-450. For more, please see .

I also think Foldit players should be able to replace white hydrogen atoms with green atoms.
This can be done via Free-radical halogenation, which is replacing a hydrogen in a C-H bond with a halogen atom via a diatomic halogen molecule. For more, please see .

I also think Foldit players should be able to replace a green atom in a Carbon-Green atom bond with a carbon atom triple bonded to a blue atom (nitrile group).
This can be done via nucleophilic substitution. For more, please see .

Also, Foldit players should be able to replace the nitrile group in a Carbon-Nitrile bond with either:

  • a carbon connected to two red atoms (carboxylic group - the top of glutamate), via "hydrolysis under reflux with mineral acid catalyst, or"
  • a carbon single-bonded to a blue atom (amine group - the top of lysine), via "reflux under dry ether, followed by addition to H20"

For more about the possible reactions with Nitrile groups, please see .

There are many more chemicals and chemical reactions that involve organic compounds in organic chemistry, so the premise of having the limited options as provided in the Reaction Design tool as of now for designing a ligand that can bind to a receptor is not, even intuitively, efficient nor creative in the long-term. I think we are not looking at more efficient and strategic ways to do things like this.

I think this contextual menu would be helpful in designing ligands because it would allow for the strategy to make small atom-by-atom modifications on the ligand and then wiggling, which stimulates the process of natural selection/evolution, like mutations in DNA. Also, it would be helpful in terms of diversity and creativity in the designed ligands.

In this feedback would also like to take into consideration and respond to the premise that "the last thing we want is to create a wonderfully scoring small molecule that wouldn’t be possible in the real world, or worse would explode!". If that is the case, the preventative measures that should be taken would be to make a referenceable list of all the criteria that should be met for the ligands, and what can and can not be allowed. It is just like coding.

You have to outline all of the possibilities and things for AI to truly develop an efficient and outstanding way in developing solutions to problems.

11. Another thing I want to point out for the developers and scientists to consider that fits with this feedback in general is the possibility of having the feature of calculating the energy & compounds needed (chemicals that are only needed to make this particular ligand) & cost(as in money) and scientific equipment (e.g. enzymes, yeast cells, centrifuge, etc.) needed to add these functional groups to the ligand in this supposed contextual menu, and displaying it to the Foldit players and possibly adding any criteria (e.g. cost limit, energy limit, equipment not allowed, etc.).

I believe the consideration of the situation that would occur on the scientist's side of the table would increase the efficiency and productivity of the Foldit players, because the more goals they have to reach, the more motivated they are. I also believe that the premise of having optional requirements of utilizing certain chemical reactions that can be conducted with certain scientific equipment can be energy-efficient. Plus, with these requirements, the planning will already be half-done to make this compound. Feel free to modify this idea to fit your specialties as needed.

Thank you for reading through my thoughts on nspc's idea!

From, donuts554.

jeff101's picture
User offline. Last seen 1 day 2 hours ago. Offline
Joined: 04/20/2012
Groups: Go Science
8. In the tool, let the left and right arrow keys go left and right through rows,
and let the up and down arrow keys go up and down through columns. This way, we
can count more easily which item we are selecting in each row and column.

9. Make sure that the order you select things in the rows and columns 
does not affect the final outcome.

10. In Puzzle 1855, it would help if the label for the ligand (residue 290) 
would list how many atoms were in the ligand and which row and column #'s 
were used to design it. For 1855, the numbering could be T1 for the far-left 
column in the top row, L1 for the top row in the left-most column, L6 for
the bottom row in the left-most column, M1 for the top row in the middle
column, M15 for the bottom row in the middle column, R1 for the top row in 
the right-most column, and R2 for the bottom row in the right-most column. 
Thus, the label for residue 290 could be T1 L3 M5 R2 (the ligand that gives 
1855 the score -7781.683) or T1 L6 M12 R1 (the ligand that gives 1855 the 
score -17611.134). If my scores disagree with others' scores, please let me 
robgee's picture
User offline. Last seen 3 days 23 hours ago. Offline
Joined: 07/26/2013

Please add some kind of identifier for the bases, reagents.

jeff101's picture
User offline. Last seen 1 day 2 hours ago. Offline
Joined: 04/20/2012
Groups: Go Science

In one corner of each base or reagent picture, you could put a number
to help identify it. You could also put its chemical abbreviation,
like CH3CH2OH for ethanol.

Also, after the ligand is formed, you could add a label to what
appears when you Tab on the ligand, and this label could give
the ligand's whole chemical formula, its total number of atoms,
and which row & column numbers were selected to make the ligand.

jeff101's picture
User offline. Last seen 1 day 2 hours ago. Offline
Joined: 04/20/2012
Groups: Go Science
In one corner of each base or reagent 
picture in the Reaction Design Tool, 
you could put the chemical's name, 
like acetic acid for CH3COOH. This 
would be educational at least.
agcohn821's picture
User offline. Last seen 3 hours 50 min ago. Offline
Joined: 11/05/2019
Groups: Foldit Staff

Hey everyone!

Thank you so much for your suggestions and input--it has been passed along!

LociOiling's picture
User offline. Last seen 1 hour 27 min ago. Offline
Joined: 12/27/2012

The current reaction design tool is described in Reaction design puzzle and Puzzle 1855 reaction design on the wiki.

Based on that effort, here are a few additions to the other suggestions on this thread:

  • The highlighting of the current base and reagents is very hard to see, since it's close in color to some of the GUI elements. A hot color or pattern would make it easier to see what's selected. Just replacing the default color (71 131 131) with red (255 0 0) worked for me.
  • When the reaction design panel is opened, the same items are always selected. For 1855, it's the base plus the top item in each of the three columns. I'll second Susume's suggestion 2, the panel should indicate the current picks in some way.
  • The default selections in the panel are not what's shown in the glowing preview image. While the top items in each column are highlighted, the preview shows items 4, 4, and 2. Super confusing.
  • Clicking the "x" in the upper right cancels the dialog without changes. Having an explicit "cancel" button would be clearer and maybe a bit more consistent with other tools.
  • The "accept reaction result" button isn't labelled. Just an "OK" label would do, and maybe make it a check mark instead of lab ware.
  • In the selection interface, the ligand must be selected to use the reaction design tool. The selection highlighting gives a Michelin Man effect, making it difficult to see the preview image.
  • In the selection interface, if the move tool is active, it's shown on the ligand while the reaction design panel is open, but it can't be used. Dragging on the background can rotate the preview, but it's kind of awkward. It'd be nice to either have the move tool work or hide it.
nspc's picture
User offline. Last seen 10 hours 20 min ago. Offline
Joined: 03/26/2020
Groups: Go Science

This kind of interaction I found in a old Foldit video, worked fine.


Developed by: UW Center for Game Science, UW Institute for Protein Design, Northeastern University, Vanderbilt University Meiler Lab, UC Davis
Supported by: DARPA, NSF, NIH, HHMI, Amazon, Microsoft, Adobe, Boehringer Ingelheim, RosettaCommons