12 replies [Last post]
Joined: 05/19/2017
Groups: None

I was curious to know how the Tutorial revamp is going. I'm especially interested in what kinds of changes are being made to be friendlier to a wider audience, and whether or not they're the kinds of changes that could potentially make this game appealing to my varied gamer friends, even if it's only for a short while.

There's quite a bit that I've learned about this game through streaming and having chatters like Susume and Tokens give me crash courses in things like the Ramachandran Map dragging. The problem is that these are things that I would not have learned from the Tutorial levels and normal gameplay alone, and implicitly forcing players to go wiki-diving for learning is a very easy way to lose impatient players who aren't having fun with the game already. You could make the case that Minecraft was/still is a wiki-diving required game to learn but having said this it is much more of a free-form intuitive sandbox game than a scientific, very niche protein sandbox. A good tutorial revamp will have to cover a lot of these commonly used advanced tactics that folders use on a daily basis before running their scripts (and don't get me started on how intimidating those are to a new player either, phew. half the time nobody really takes the time to write down what their script does in plain words).

Teaching players how to play the game within the game is a hallmark of good game design. A prime example of this is how stages in Mega Man present obstacles in a safer environment so that the player learns how they behave before they proceed onwards to more dangerous instances of the same obstacle. Familiar, yet different. Puzzles in critically acclaimed puzzle games such as Portal also follow this pattern, as well as puzzles in non-puzzle games like the ones in Legend of Zelda Dungeons. By this same vein, it'd be nice to have multiple puzzles for practicing the different techniques presented so as to build mastery before plunging into the Beginner Puzzles and Weeklies.

When the new tutorial does come out, I will most certainly livestream myself playing through it and hopefully draw in a few audience members outside the fold (pun intended) to see what's going on. Hope this discussion helps!

smortier's picture
User offline. Last seen 4 days 6 hours ago. Offline
Joined: 03/10/2016
Groups: None
Thanks so much for your comments, S0ck.

I wholeheartedly agree with your statements. Thorough and fun tutorials set the groundwork for fantastic gameplay (Half-Life, anyone?!?).

I've emailed the admin team to see if the person working on the tutorial changes can hop on and give an informed update.

Thanks, again. Really thankful for your perspective and the time you commit to communicating that perspective.

Uttkarsh's picture
User offline. Last seen 2 hours 17 min ago. Offline
Joined: 05/03/2017
Groups: None
New Tutorials.

Hey S0ckrates,

Sorry for the late reply, but yeah we have a few tutorial levels coming up which will delve into Foldit tools which weren't touched in the tutorials till now. We are always looking to make our tutorial levels more inviting and accessible to new players, that is integral to any good introductory level. Your words very well explain how intro levels should behave and we hope to provide new players a better first experience with Foldit. :)

That be awesome if you'd play through the new levels on live stream when they come out. Thank you for your continued support

joshmiller's picture
User offline. Last seen 13 hours 6 min ago. Offline
Joined: 09/08/2017
Groups: None
Hey S0ckrates, If you would

Hey S0ckrates,

If you would like to help the tutorial revamping, I would like to know what strategies you've figured out as an expert folder. If you could give a few tips to new players to solve puzzles better, what would they be?

Thanks :)

Joined: 05/19/2017
Groups: None
I'm no expert, but...

The following list is either techniques that I've learned or techniques that I want to learn as a semi-experienced but not quite consistent top 10 leaderboard folder.

- Different puzzle types call for different strategies to approach. I personally have the most experience with De-novo folding since those are the easiest to grapple with for me, but having tips on big picture approaches for the puzzle genres is the first thing. Explain what is "early game, mid-game, late game" in Foldit. For example, in the early game for De-novo puzzles, sometimes it's useful to take the predicted structures, use the cut tool to separate them, and lay them out in front of you before choosing what goes where and how the protein is going to be packed. In particular, the Black Belt Folding series is something I think could be translated into an in-game tutorial about general strategy rather than individual mechanics. It's one thing to know how to do the individual actions in the game, it's another thing entirely to know how to put them to use without getting overwhelmed with the complexity. P.S. on this one: I need Design tips other than mutate everything, wiggle, and hope for the best. There's got to be a better workflow than that.

- Use cut/rebind! It turns the amino acid chain into more manageable chunks, kind of like molecular legos (moleculegos if you will) so that you can play around with them and figure out where things go. Moving and wiggling individual segments is taught to you in the old tutorial, but never really put into practice with further examples, so find your own opportunites to do so in the De-Novo beginner puzzles.

- Don't be afraid to assign your own secondary structures on unknowns! The whole point is to see if humans folders can get creative solutions faster than a computer algorithm would, so if intuition strikes, go for it!

- Try recipes for sweet sweet score optimization. You're probably not gonna know what they do so it's probably wise to ask in global chat since the recipe authors don't really put what the recipe actually does in the description and instead just plaster patch notes and expect everyone to pass it on by word of mouth. Annoying, I know, but community learning is community learning.

- Don't forget to use a Medium wiggle at clashing importance (or CI, if you haven't learned that abbreviation by now) 1.00 so that you have the absolute best score possible. Medium wiggle forces backbone ideality calculations when it works, so watch those points go up when you do it. Medium implies there would be a "high" wiggle power, but there isn't. Don't ask me why, I have no idea.

- If you wanna bend an individual segment, try Ramachandran map dragging. This technique (unfortunately) takes a little UI menuing since it's buried beneath some menus and it's not taught in the old tutorial, but go to selection interface, select the segment you wanna bend another way, open the Rama Map (under some menu I can't remember, again, UI is clunky as hell y'all), and look for the little white square representing the amino acid selected. Drag that around and watch how it changes; it's a bit unintuitive but probably one of the better ways to bend a specific segment if you don't think freezing, rubber banding, and wiggling will get you what you want without a denatured mess.

- This tip is also buried under obfuscated UI, but try switching the color mode in the view tab over to "Hydro". This will colorize the residues in your protein by hydrophilicity, so hydrophillics will be completely blue (even the backbone) and hydrophobics will be, you guessed it, orange. This can help a lot in the early game when you're trying to figure out what packs where in your solution.

- Remember to right click your alpha helices and hit "Idealize SS" if they're all sprawled out. They won't automatically wind up when you hit wiggle, much to your chagrin.

- If there are 2 cysteines (short, orange hydrophobic amino acids with yellow tips denoting the sulfur on them), more often than not they're going to form sulfur bridges. Band them together, set the length to about 2.00 by right clicking the band, and crank the strength way up. Wiggle, or use cutting to bring the pieces together. This example doesn't come up often, so if this could be a recurring example that players face to learn to recognize these opportunities, that'd be great.

In general:
The tutorial needs to be expanded with more puzzles to hone mastery of individual techniques in conjunction with other techniques through practical examples. Mere tips are not enough: Opportunities to practice said tips are the key to designing a good tutorial (or the rest of the game, for that matter) that onboards new players and gives them all the tools they need to discover creative solutions. Not to say that hidden mechanics are fun to discover, but with a game as esoteric and arcane as this, clarity and competence are the key things players need to feel as they get acquainted with folding.

joshmiller's picture
User offline. Last seen 13 hours 6 min ago. Offline
Joined: 09/08/2017
Groups: None

S0ck, this is super helpful, thank you so much. If you think of any more, feel free to email me directly at miller.josh@husky.neu.edu. I'm going to be doing a lot of the bulk work on revamping the tutorial's UI and implementing more principles of good game design over the next few months, although it may be a while before we push out the changes.

smortier's picture
User offline. Last seen 4 days 6 hours ago. Offline
Joined: 03/10/2016
Groups: None
Drafts of the New Tutorials are Up in DEVPREV

We'd very much welcome your perspective on what's been done. Feel free to shoot me a message with any thoughts you have. I'll be compiling feedback and passing it onto the student who has been working on them.

Susume's picture
User offline. Last seen 2 hours 20 min ago. Offline
Joined: 10/02/2011
New default recipes?

While you are revamping the tutorials, would you consider replacing the existing 4 default recipes in the cookbook with scripts that do something more useful? I suggest Blue Fuse as one, and a version of Zero Length Bands as another (maybe with a little revamp of the starting dialogs). One of the Cut and Wiggle scripts might make a good third.

If you are determined to have a sample GUI recipe, Blue Fuse can be done in GUI.

Joined: 05/19/2017
Groups: None
Second this.

This handful is more representative of the typical Recipes used in the uh...I can't believe I'm saying this for this game..."meta." Yep. Foldit metagame exists. Time for newbies to learn the meta.

joshmiller's picture
User offline. Last seen 13 hours 6 min ago. Offline
Joined: 09/08/2017
Groups: None
Hah, every game gets a

Hah, every game gets a metagame eventually. This is good feedback, thanks. Should be relatively easy to do, I'll look into it when I have some time. While I'm here, I have a few more questions that could be useful for working on the tutorial:

- You mentioned earlier that there's an "early game", "mid game", and "late game". What are they and what general strategies work for each section? (I imagine they depend on the puzzle type, though)

- For the more specific tools (changing weights of rubber bands, using different view modes, wiggling individual segments), how do you know which tool to use when? What scenarios are each tool best at?

- What are the most confusing parts of the UI?

Joined: 05/19/2017
Groups: None
Early/Mid/Late Game

Everyone's gonna have a different natural progression of things for Early/Mid/Late game, and I can't speak for all players, but I'd imagine this is a good layout. Any script-heavy/script-only or hand-fold only players wanna weigh in, feel free.

Early game = Mainly hand folding, and choosing what kind of solution you wanna go for. Every puzzle is different and contextual, so for the Aflatoxin puzzle for example I'd be manually mutating residues to hydrogen bond or pack. With de novo freestyle (structure prediction) puzzles, early game is probably picking and choosing to keep/replace secondary structures and moving them into place. And as for monomer design puzzles, those early games are like de novos except you can do whatever you please to the residues. Shake, wiggle, mutate as needed. Script use is pretty limited, with the exception of quality of life scripts to do things like spot cysteines in some puzzles and band them together.

Mid game = Most of the big stuff is done, so it's time to do specific refinements. Maybe you want a specific residue to be somewhere, or perhaps you wanna re-do a secondary structure placement but not upend the whole thing. Still hand folding, but none of the from-scratch stuff. May include recipes, this area's a blurred line. The scripts/recipes that could be used here aren't really "slow churners", but really "one-off" utility recipes that do one pass and clean things up. These include single pass scripts like Blue Fuse or Fast Relax, or any of the previously mentioned quality of life scripts, like Bridge Wiggle to band cysteines together. Quickfix is another good one too. The whole point of this phase is to get the protein as best as possible by hand (more or less) in preparation for late game

Late game = Slow churner scripts. We're talking the works: TvdL's, Random Idealize, AFK3 (BounceWiggle), BandFuze, the list goes on. This is where you set up these scripts in their respective settings windows, tell it what not to mess with if applicable, and then set it and forget it. This is what we have people leaving their computer and game on for hours for: brute forcing more point refinements out of the existing structure. At this point, major edits are probably padlocked, or would get wiggled/mutate pass'd out. I'm less familiar with the metagame for this phase, but it's pretty easy to let a script do it's thing if it's simple enough to use. Don't forget a final Medium/High power wiggle once your scripts are all done to force ideality calculations for those last few partial points of score! Oh, and did I mention "multiboxing" is a thing? You can have a late game fold churning in one instance of the game while hand-folding another solution candidate in another instance of the game, assuming your computer can handle it and your game won't lag so much it becomes a PowerPoint Presentation.

Joined: 05/19/2017
Groups: None
Tools of the Trade

What to use where is dependent on what you want to do with your solution. Mastering the controls is key to being able to do what you want in this game. I can't name everything off the top of my head worth knowing but if you bring something up like in the parenthesesed list you got there, I can prattle off what I know. There's a lot to teach here, and I shouldn't be the only one to teach it, but I can explain what I know in straightforward English hopefully.

Band strength: Sometimes, you really really want your band points to be at that specific distance you set, maybe to force a hydrogen bond, or whatever. Crank band strength up! In truth, scripts actually have the ability to set band strength beyond 1.50, but for whatever reason players can't do that. I guess with great power comes great responsibility. As for weaker bands, I'd only use them if wiggling or pulling with the bands active completely wrecks the entire setup in to something so mangled it's useless.

Band length: So it turns out the length units are in angstroms, if I remember what Sus' told me correctly. Hitting "use current" will give you the current length between the two band points, so you can crank it up or down as needed. Which should lead into a sometimes useful discovery: bands can push too! If something's too close, set band length farther than current, and the band will start to repel the points away. Vice versa if you want something even closer than default. Also important to remember: Hydrogen bonds will occur somewhere between 1.5 and 2.99 usually, if you're trying to use bands to help out. Check 'em out.

View modes: The view options are worth a post in their entirety.
- Toggle Settings
> Toggling voids off is a good one, since they usually get in the way more than they honestly help (unless your score breakdown by mousing over + hitting tab with certain residues is suffering from packing. That reminds me: global score breakdown when? I don't have time to add up all the individual categories from the tab tooltip.).
> Show backbone issues is something that's usually always on for me starting in mid-game: if there's a major problem, it becomes way easier to see.
> Toggling bonds on and off can help ease the lag and also help you focus on the type of H-bonds you're seeking to form or monitor.
> Show bondable atoms is a godsend. It opens up the sort of problem solving reflexes you get with jigsaw puzzles. Seeing the bondable atoms on the backbone/sidechain is a good way to figure out whether or not it's worth losing a few points to get an extra bond.
> Showing constraints is usually a good idea, since they're exactly what they say on the tin.
> Showing expected residue burials, isosurface, symmetric chains, outlines are usually not a thing I end up using. Symmetry puzzles haven't been around for me to play, the isosurface doesn't always work and lags like hell without really providing much, and outlines are personal preference.
> Dark background. No eyebleed.
> Turning off pulse while working can aid performance, but also having it on is useful to see where your script or wiggle is working.
> X-ray tunnel for ligand can help you find your bearings if your ligand suddenly gets buried behind protein.
> I personally play with hide/fade GUI off.

- Color coded for your chromatic convenience: In the Color menu there are a few worth noting.
> EnzDes makes it easier to see ligands, like in the Aflatoxin puzzle, to get your bearings.
> Hydro is useful for really getting used to pointing orange hydrophobics in and blue hydrophillics out since it colors the backbone blue/orange too. Great for helices!
> Rainbow isn't only just pretty to look at, it's a natural progression of ROYGBIV from one end of the protein to the other, so you can find the middle and ends really quickly.
> Finally, Score/Hydro+CPK is my go-to default: It looks just like standard Foldit with sidechains colored blue/orange and backbones colored based on score. CPK just makes it so that polar side chains have their polar H-bonding bits visible even when "Show bondable atoms" is turned off.

- View Protein Menu
> Cartoon is the default, with cartoon ligand and cartoon thin shaking things up only slightly. The easiest model option to work with, and general purpose too.
> Sphere, I've heard according to Susume, is useful to see whether or not a hydrophobic is exposed? It's pretty bulky and awkward.
> Stick views, with or without hydrogens, are really nice for seeing how all the atoms are actually arranged. Organic chemists in the house?
> Line views are like their cousin stick views. Except probably nigh on invisible since they're so thin. I guess it would lag less?
> Trace Line/Tube I haven't really found much more useful than cartoon.

Wiggling individual segments: I tend to bust out selection interface to do this whenever I'm really getting into the nitty gritty specifics of a design. The most immediate example? The Aflatoxin puzzle. I'm trying for score attack, true, but I foresee a good science answer having a boatload of hydrogen bonds at the cost of some score. So, I'd mutate a new residue and try to pull it towards a donor/acceptor. One problem though: it is SUPER hard to get a sidechain exactly where you want it to go. Wiggle usually fixes the difference, but sometimes you don't wanna move the rest of the protein, you just wanna make that one segment find it's comfy spot. Selection interface, select, wiggle. Done.

Joined: 05/19/2017
Groups: None
UI Gripes

- Selection interface completely overhauls the familiar UI which was a little clunky and obfuscated to begin with. I don't expect any newbies to not be scared off from this without a walkthrough or some sort of consistency/parity with the original interface.

- The original interface is pretty outdated, and could probably do with a merging of the newer "sidebar menu" style so common to a lot of apps and games and such. Also, menus must be collapsed whereas in selection interface everything is cordoned off in a draggable window that can be kept on hand.

- Resizing elements of the UI is inconsistent and unavailable for some elements.

- The fact that there's a whole bunch of commonly used stuff in the 'meta' hidden behind the Advanced GUI toggle in the options menu leads me to question why it's hidden by that toggle option in the first place.

- Differentiate shake/mutate/wiggle icons. Shake messes with sidechains, wiggle with mess with everything. Wiggle backbone/sidechains aren't used as much.

- Only biochemist smart folk really know what the Rama(chandran) Map is. Even less, know how to use it to rotate a single segment as a hinge, which isn't really possible in the same way with dragging.

- Cut>move>rebind is really useful, but barely communicated. I forget that it's there half the time.

- Social tab is pretty much pointless at the moment.

- Other than the leaderboards, I was confused that people kept mentioning some sort of progression score in the game when I realized it was referring to global score which is only viewable on the main website. No in-game feedback for that sort of thing.


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, Microsoft, Adobe, RosettaCommons