Hand Fold Proteins

Case number:699969-995442
Topic:General
Opened by:karstenw
Status:Open
Type:Suggestion
Opened on:Tuesday, July 2, 2013 - 16:28
Last modified:Saturday, July 20, 2013 - 21:02

I was thinking that there really is no reason that we couldn't get the regular version of a protein at the same time that we get the hand fold only version. This could be nice because we can test our own hand fold immediately.

(Tue, 07/02/2013 - 16:28  |  32 comments)


bertro's picture
User offline. Last seen 12 hours 31 min ago. Offline
Joined: 05/02/2011
Groups: Beta Folders

I have to disagree because the regular version will "influence" the hand-folded one TOO much in my opinion since the regular one will certainly give a better score each time.

Joined: 09/21/2011
Groups: Void Crushers

In my opinion (now being nr 1 at handfolding and still hating it) handfolding should be replaced by non sharing and no picture sharing but allowing all the scripts.

MurloW's picture
User offline. Last seen 12 weeks 1 day ago. Offline
Joined: 11/21/2012

I can totally see why they want us to handfold first, there's bound to be greater diversity and it's a greater challenge for us having to work out little kinks ourselves. They shouldn't give us both puzzles at the same time, imho, it would even further discourage playing the handfolds. What they should do is shorten the timespan on the handfold, atm we're forced to put serious mindnumbing rebuild and fuzing work into them if we want a respectful rank. If they were only, say 3 days long, we could just throw a bunch of configurations together, see what kinda scores and work on those with scripts later on. That's what I already do. I think that would also give less experienced players a chance to get a good rank, instead of only the folders with end-game knowledge.

Also, I don't think *all* scripts should be allowed, but some tiny things should.
Like setting all bands to their current length, it's a great pain to do manually when u have a lot of bands, and there's no easier way on handfolds. Or information gathering recipes, like jpred helper, or AA Copy Paste Compare.

brow42's picture
User offline. Last seen 1 week 3 days ago. Offline
Joined: 09/19/2011
Groups: None

Yes, I would be interested in seeing LUA 2 scripts, with "do-er" functions wiggle, shake, and rebuild, idealize (and tweak if it exists) disabled.

Joined: 09/21/2011
Groups: Void Crushers

Second this. Said it before.

Joined: 04/15/2012
Groups: Beta Folders

You have responded non this, but for those who don't know, I submitted a feedback to have change band length included: fold.it/portal/node/995038

marie_s's picture
User offline. Last seen 2 years 2 weeks ago. Offline
Joined: 05/18/2008
Groups: None

I had a respectful rank in 737 without many fuzing, walking rebuild, put many bands or gui scripts.
I just rebuild (heavily on some parts, shake, wiggle sidechains, tweak a little, flip a sheet always on part always at ci=1. I choose the parts to work on.
I stopped to work on it some days before the end because I have no idea for another fold.
The timespan could be shorter but it is good to have a week end in it.
I think no scripts should be allowed.
After 5 years here, I think that the real pleasure will be if my fold is close to the native when it will be released.
I think the hand folding should stay for the pleasure of the players who like them like me and do them without been bored a second.

gitwut's picture
User offline. Last seen 45 weeks 5 days ago. Offline
Joined: 05/18/2012
Groups: Contenders

I definitely agree with shortening the duration for the hand folding round, though everything else (no LUA scripting nor sharing) are fine as they are.

bertro's picture
User offline. Last seen 12 hours 31 min ago. Offline
Joined: 05/02/2011
Groups: Beta Folders

Remember anyone can capture a screen or window on a computer and share it if they want.

So it NOT possible to stop picture sharing at all.

Joined: 11/03/2011

In my opinion,hand folding is a complete loss of time.
The diversity argument is weak,because there are more messy solutions than promising ones.
Furthermore you can start the next client where scripts are authorized from scratch and
it may be safer to do so.
I never understood the argument saying "we are not able to select the best solutions
which give the best rmsd on CASP so we want to have more diversity in the solutions"
How on earth does this help you selecting the best solutions with rmsd criterium????
There is an analysis to do to check from where comes the discrepancy between the best
foldit solutions and the rmsd criterium breaking down the score components in a variance analysis.
.If i had to make an educated guess,i would look
in the sidechains configurations.many recipes use wiggle all command which leave lots of poins
that a wiggle sidechains would have found.Finally lot of work has been done on wiggle backbone
and little on wiggle sidechains.obviously I may be wrong,and my view is biased as i hate
hand folding and am not good at it.

Joined: 02/08/2012
Groups: None

As a fellow hand-folding lover, I have to disagree with this. It contradicts the purpose of hand-folding. I think hand-folding is fine just the way it is.

Susume's picture
User offline. Last seen 1 day 20 hours ago. Offline
Joined: 10/02/2011

I believe the expected benefit of a hand folding round is to reduce false positives. A false positive is when the foldit score says a solution is good, but really it does not match the native well. One way to get a "positive" result (a good foldit score) is to match the native well; another way is to create any plausible protein (orange in, blue out, compact, etc.) and run powerful scripts against it for a long time-- even if it doesn't match the native. This second way is (mostly) prevented in a hand folding round, so there is a better chance that the best scoring solutions actually resemble the native.

One down side of the hand folding rounds, for the players, is that the software still rewards the kind of repetitive actions that scripts do. So some of the top scores will go to players who have the time and inclination to manually imitate scripts (especially rebuild scripts) - in other words, to grind. Grinding is mindless and boring -- that's why we prefer to let scripts do it. Sadly, rewarding grinding may be the necessary price of reducing false positives.

spmm's picture
User offline. Last seen 45 weeks 2 days ago. Offline
Joined: 08/05/2010
Groups: Void Crushers

I really like hand folding, even with just the few GUI scripts we have (thanks SP), a more level playing field for folders without powerful computers, great coding skills or as has been mentioned 'script game skills', also having a week to do them means you can take a break and try lots of versions.

I appreciate being able to take my time and learn how to hand fold, it was an opportunity many people asked for because those with good hand folding skills had developed certain advantages even against scripts. They all moaned about how hard it was though, so having GUI scripts is great.

Joined: 09/21/2011

well, this is an interesting topic. Anyway, how about an overlap? if a regular protein would normally run for 6 days, what if handfold runs for 5 days, but on day 3 the regular protein is released for 3 days. its a good compromise that stops hand folders from grinding away manually (which is a ridiculous waste of time) and it shortens the time that script users can run scripts, forcing them to load their handwork in for scripting, rather than running a script only protein (which i find ludicrous to do anyway). We might be better off with 50 points per puzzle, so those of us who score well on handwork aren't rewarded twice for the same pose. I had r2 for 738 (got tired of grinding so i quit), and now i have r1 for 743.

drjr's picture
User offline. Last seen 22 weeks 1 day ago. Offline
Joined: 06/27/2009

If the local wiggle worked like it used to, the handfolding rounds would be great. As it is, I find them tedious. The low backbone scores are hard to get rid of and take so much time to deal with.

The original topic: to run them concurrently? Better to ditch the hand-fold.

spmm's picture
User offline. Last seen 45 weeks 2 days ago. Offline
Joined: 08/05/2010
Groups: Void Crushers

agree with drjr about the local wiggle and low backbone scores - but if you really don't like hand folding don't do it. Releasing a sharing round with scripts at the same time defeats the purpose imo.

spmm's picture
User offline. Last seen 45 weeks 2 days ago. Offline
Joined: 08/05/2010
Groups: Void Crushers

Actually why not run the second puzzles after the freehand, ie the one with sharing and scripts just for 3 or four days not six? Scripts should get you there faster.

Joined: 09/21/2011

Yes, that's what I was thinking spmm. If the second one ran for like three days.
But I was thinking It could be released with an overlap of two or three days with the handfold puzzle. Or maybe only release one puzzle, but with scripts locked out for a period of time? I don't know.

Joined: 09/21/2011

Let's be clear. "Hand folding" is only useful in setting up a pose for that protein. Most of us can set up several poses within a matter of an hour, or two. The rest of the time is spent "hand scripting" which is not the same as "hand folding". Hand scripting is an effort to condition the protein to accept the pose that you designed in that first hour, or two. The ulterior motive for those who prefer hand scripting is that they believe they can handicap other players with fast computers (which is highly over stated) and thus "even the playing field". This is silly because the scoring is in the pose, and the days given are plenty for even a slow pc to check their pose with scripts. The real reason for hand folding is two-fold. Cheat scripts are foreseen and the only way around that is hand folding at this time; and scripting allows people who don't understand folding to run dozens of random useless script only proteins where one or two somehow manage to score well even though they are no where near a native pose. The "false positives" are getting to be a pain in the neck, I suspect. This is why we need to figure out how to make this work without the opinions of the hand scriptors and the script-only people from muddying the waters. In my opinion, running the protein twice and having a hand only version is a very poor gaming solution. Those who can't sit at their computer 24/7 hand scripting (most humans) are going to see their global rank plummet. Crowd sourcing becomes a useless goal too, "hey, come tie yourself to a computer for hours of enedless repetition! yay!" get serious.

marie_s's picture
User offline. Last seen 2 years 2 weeks ago. Offline
Joined: 05/18/2008
Groups: None

Sorry Karstenw, but your description is not at all my experience.
Try to think that other can thought and act in a different way than you.
I begin to fold before recipes.

I don't find a pose in hour and try after that to bind the protein in the shape I think the protein could have.
I find a pose by working slowly on the protein, choosing every step by watching the protein.
I almost never a move by random or for all the protein...: I never hand script on hand fording puzzle.

My goal is not global score, my goal is the fun of the game.

I am sorry than players who begin after the recipes cannot take fun like me in hand folding and don't saw how they could learn dealing with the protein.

Joined: 11/05/2010
Groups: None

I can't begin to describe how negatively I feel about hand-folding however every once in a while it comes in handy for short periods of time to reduce issues on a protein. Many use it to create solutions that the human eye finds appealing. Luckily, most of the time, nature, by trial and error ( variation ), creates a solution that isn't beautiful to the human eye. I find those much more appealing then any hand-folded protein I have yet seen. Nature is much more wonderful then what we can conceive and for all intents and purposes has few limits to what it can do biologically speaking. To limit ourselves to what WE find appealing by the hand-folding method is the height of human self-ego. Computers and nature both provide a means to test a nearly unlimited # of variations, given enough time. I am not an artist, nor will I ever be. I am simply an old man technician who knows how to keep his computer running well enough and long enough to test many ( but not nearly enough ) variations. AND none of that would be possible without powerful recipes and we need more, much more, of them. The one thing I wish for constantly is the ability to write recipes. Those of you who can are blessed with insight that rivals any artist. There is nothing wrong with repetition, else there wouldn't be 6 billion of us and the diversity of life on this planet would be much less then we consider appropriate or normal. Nature is doing something right. I want my computer to be doing something right also, even if I am over across town or half a world away. Otherwise this computer need not be wasting electricity and heat dissipation on a project that without the use of recipes is nothing but a canvas for some particular artist to draw their dreams and imaginations on. To much written, I know. Sorry about that. Recipes keeps Foldit current, Hand-folding, except in short bursts, is a regression back to earliest stages of Foldit. A place I have been, and never want to be again.

Joined: 09/21/2011

To Marie. If you enjoy trying to predict your position while ignoring the score of a better rebuild position in order keep the fold position that you personally chose then I applaud you. And by your own admission you are not playing for score,so you will continue to handfold while ignoring the score and do not need handfold only puzzles to do so.
To Hanto.At this time we do not have an understanding of what beauty is, but we do know when proteins are piling up that look nothing like the native.

Joined: 05/26/2008
Groups: Hold My Beer

Personally I love hand folding, it holds my interest much longer. When I can just throw a script at something and walk away, I usually do.

I think that the focus should not be to change hand folding, but to increase the effectiveness of GUI scripts. There is still a great deal of unexplored ground that can be covered in that regard to make these script types far more effective than they currently are.

Joined: 09/21/2011
Groups: Void Crushers

If handfolding is there to stay, here my top 3 wishlist for GUI functions:
1) Make it possible to run GUI scripts in the background WITHOUT having to tweak the options file but with the speed increase we get when we run LUA scripts in the background.
2) Add a function to make zero length bands
3) Add a function to make distance preserving bands

Joined: 05/26/2008
Groups: Hold My Beer

I definitely agree with more control on bands, the ability to add opposing bands would spawn a whole new subtype of GUI recipes. I have two scripts I tried making which I simply cannot make work without band length control. (as I need things to pull apart not together) I could make them in LUA, but that's a lot more work as I am not proficient in it and don't really want to take the time to make them currently.

What do you mean by distance preserving ?

Joined: 09/21/2011
Groups: Void Crushers

Distance preserving is when the band target length is equal to the distance between segments. My bandmaker script (which I use all the time for stabelizing) uses those. It prevents the protein flying apart while giving more freedom as zero length bands.

Mark-'s picture
User offline. Last seen 38 weeks 3 days ago. Offline
Joined: 08/05/2009
Groups: Contenders

Wow! Lots of negativity here.

If you you don't like hand folding, don't do it. Leave it to those of us who do like doing it. That is one of the reasons that categories were were introduced, so that you didn't have to play every puzzle. Concentrate your effort instead on where you think it will do most good.

I personally don't like puzzles where scripts are needed to get anywhere, so I don't play them. My cpu's and graphics card are put to work on other projects using BOINC. Rosetta@home is always happy to use your spare computing power ;)

Joined: 09/21/2011

Yes, if gui is improved, then we don't need to run the proteins twice. That could work well I think. However, if we continue to release the same protein twice then I think I have a solution.
Of the 2 releases:
The game should only accept your single best rank for scoring toward the global ranking. This would allow the handfolders to get a decent rank without touching the regular protein, and it also allows those who do scripts to get a decent rank without touching the handfolding. Also this allows those of us who often like to run both proteins to compete against themselves without being twice rewarded/punished for the same protein. And it might provide good data to the devs. And as a bonus, the statement "if you don't like it, don't do it." actually becomes a reasonable sentence that doesn't make my skin crawl every time I hear it.

Joined: 09/21/2011
Groups: Void Crushers

I really like this suggestion, if the best rank is used for all categories of the puzzle except the handfolding of course. I probably will play the handfolding part just to get the feel of the puzzle but will not waste so much time refining it as I do now.
And I fully agree with your last sentence.

spmm's picture
User offline. Last seen 45 weeks 2 days ago. Offline
Joined: 08/05/2010
Groups: Void Crushers

so do you want to do that with every other repeated puzzle in foldit? there are a lot of them. Same logic applies.

Joined: 09/21/2011

i think i need to reword the question. "karsten, if a protein was released 8 times in the same 4 month window and you were able to roll your solution over through all 8 proteins would you would ok with that right? after all, we do repeat proteins around here, so you then must follow the logic wherever it takes you. otherwise you are not consistent." You caught me...if we get 8 proteins and i manage to stay in the top r5 for all 8 proteins, i'll still cry foul from a gaming perspective. even though i'm not trying to stamp out repeating proteins entirely. ok this was fun, but i need to stop, i've exhausted all my arguments i think. thanks for the challenging conversation everyone. any last remarks will be read, but i won't respond anymore. i've cluttered the boards too much already.

wisky's picture
User offline. Last seen 13 weeks 1 day ago. Offline
Joined: 07/13/2011

Lots of good suggestions in here. I would really be interested in being able to actually see the native structures for a lot of these puzzles after the fact, and compare my individual fold with the native. I'm in the process of doing my first decent looking hand fold puzzle, thanks in large part to advice marie_s has given. At the end of this set for denovo 27, I would like to be able to compare my structure to the native structure (loaded as a guide) of the protein, and see where I could use improvements on my folding process/structure prediction capabilities.

Since the new wiggle came out, fuzes don't appear to working nearly as well, and they take longer. I've had to cut down considerably on the amount of fuzing I do in hand folding and non-hand folding puzzles because of this, and have started trying out new methods. My structures have also been consistently less chaotic looking, and more compact since new wiggle came out...

One possible way to account for the issue of structures not coming close to the native is a separate score/rank based on how close your structure is to the native, or even a score term that takes this into account. I realize a lot of these proteins might not have their native structures solved yet, but quite a few of them do, and some of the natives are released after a puzzle closes. I still think that should be able to weight the scoring better. Even with hand folding, you will still get a lot of false positives. Recipe only work (powerful recipes run on powerful computers) can produce high scoring false positive solutions. I know this, I've made first place on at least one puzzle (707 - http://foldit.wikia.com/wiki/Puzzle_707 ) with a false positive solution. gitwut's solution was actually considerably closer to the native fold than mine.

I know that my comment isn't necessarily about hand folding, but without a way to compare our structure to native structures, many players will have absolutely no way of knowing what better solutions look like (seeing a picture of it is NOT the same as being able to play around with it in-game) or how to create them.

@karstenw: You don't have to stop posting on this, I feel your comments have been very helpful in opening a discussion about the current state of hand folding.

Sitemap

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, RosettaCommons