11 replies [Last post]
W1zzard's picture
User offline. Last seen 10 years 50 weeks ago. Offline
Joined: 05/16/2008
Groups: None

I have play some puzzles, and most difficult is not understand how fold some parts but actually do that. I cant move or rotate some part properly, hm i cant rotate at all, or move whole part not one piece of it. Or may be it for more difficult while playing??

Rav3n_pl's picture
User offline. Last seen 11 weeks 6 days ago. Offline
Joined: 06/17/2010

Try freeze segment surrounding area you want to move. Otherwise you will drag entire protein.
Also switch to new interface, is more user friendly.

xiando's picture
User offline. Last seen 6 years 24 weeks ago. Offline
Joined: 10/19/2008
Groups: Oma Gawd
In answer


When you try to use rotate (tweak), there are a few things you must keep in mind.

1) Only helices and sheets may be rotated. Loops may be changed into sheets or helicies to allow them to be rotated, but they will not do so naturally. That's an area ripe for improvement imo.

2) A sheet or helix being rotated must have at least one (1) connected, unfrozen loop on each side.

3) A clash of significant value will stop many rotations. Sometimes the structure will simply push its way through another structure and continue the rotation, but often, when you need it to most, it will simply stop and the clash will grow until your score collapses. I'm not sure why that happens, or conversely why it sometimes simply pushes its way through, bt it does.

The same rules apply to the other "Tweak" functions, both Translate (the straight arrows) and Optimize (the balls that sit outside each rotate function)

And yes, competition-mode puzzles are often far more difficult than the tutorial levels.

Don't forget that you can always ask questions in general chat. There are often well-versed veteran players there who will provide helpful hints, although it's by no means guaranteed.

One last thing to note. Rotate works on a fixed set of index values, so one end is 0 and the other the "nth" sidechain in the structure. The reverse rotation does not reverse those values, so it will always act from one side of the structure first. (for instance, a helix will often optimize one side before the other and will repeatedly act that way no matter what you do outside tediously segregating each subsection and working it very slowly. It can be done, but sometimes things aren't "perfect".

Hopefully this clarifies some of the confusion.

best wishes

xiando's picture
User offline. Last seen 6 years 24 weeks ago. Offline
Joined: 10/19/2008
Groups: Oma Gawd

btw, a subtle solution to the problem inherent in the tweak tool is to reverse, temporarily, the indices of the entire protein chain during a reverse rotation operation. It would add a temporary complication to subscore assignments, but intrinsically, it should not introduce any error and result in an improved Tweak, since one end would not always be concentrating the beneficial aspects and the other hanging "loose", hoping for some change to come its way eventually....

for instance, suppose a right hand rotation used the default residue index values, then the reverse operation, ie, a left hand rotation, would/could auto reassign the index values in reverse order, maintaining the score function properly throughout the protein and tweak operation. Once the rotation/translation/optimize was completed, it would/could resubmit those subscores and indices to their "proper" locations in the protein string. (all that would really change is the index sequencing of the protein, ie, either 0 to N or N to 0. the scores would not change, simply the order of operation)

As I said, it's not without complication, since the value swapping would be required to maintain integrity of the valuation (obviously) but it doesn't seem an impossibility by any measure, and it might improve the functionality of the Tweak tool considerably.

...Yes, a bank of rubber bands blah blah blah and superscript GizmoThingy-4 for seven hours...I get it kids. Been there a long long time ago and got really bored. Bands have inherent problems of their own, due to their somewhat quadratic force nature, (presuming the math behind bands is as bands/springs actually work in nature) and for me at least, positioning bands in open space is a total pain.

Intuition remember?

Joined: 09/18/2009
Groups: SETI.Germany

Right, xiando, positioning bands into open space is very hard, especially, if the spatial "depth" they should be applied does matter.

saksoft2's picture
User offline. Last seen 17 weeks 1 day ago. Offline
Joined: 08/06/2010
Groups: Contenders
Sometimes after a major

Sometimes after a major manipulation, only parts of some sheets are twisted by large angles (> 180 degrees) while the rest of it has significant appeal. Other than a rebuild (which can give you a radically different random structure) or some nasty combination of "freeze + band in mid air+ pull-in-a-circle" how else can you untwist those sheets? If I freeze the good parts, I can't rotate the bad parts without seriously messing with my major structures. Sometimes, that's good. Other times, I just want to flex a sheet segment back into position and bands and pulls are not cutting it.

What I would like is a way to wring a sheet (as you might water from a rag) by fixing two rotational positions on the chain then rotating one of them. The non-selected rotator will act like a freeze by not allowing its segment to rotate, the active rotator (the one you are clicking) will attempt to rotate the partial sheet in the direction I want to rotate it.

Both sides could be active rotational sites but since they are mouse-powered, only one of them can be used at a time.

Joined: 09/18/2009
Groups: SETI.Germany

You can untwist a sheet in the new interface.
Double-click a sheet, then it should turn blue and a purple cross of arrows should appear, including a knob.
untwisting a sheetuntwisting a sheet

Pushing and holding the knob should flatten the sheet.

Joined: 09/18/2009
Groups: SETI.Germany
Wringing water

Do you mean a torsion instead of a rotation?

xiando's picture
User offline. Last seen 6 years 24 weeks ago. Offline
Joined: 10/19/2008
Groups: Oma Gawd
actually, it's more of the gravity points

to tell you the truth, it's not a spacial thing, it's the tendency of bands to auto adhere to the first object they encounter in a 2-D world. if they cross anything in paperspace, they connect. that's the pain (for me)

Joined: 09/18/2009
Groups: SETI.Germany

That's, what I meant, too.
If you want to pull a sidechain just inside the protein, into free space, but there is a visible object behind this space, the band does connect to this object (which is more far away in depth), and not to space, where you want it to have.

xiando's picture
User offline. Last seen 6 years 24 weeks ago. Offline
Joined: 10/19/2008
Groups: Oma Gawd

Yes. I recognized that issue when I first used a band on a non tutorial protein back in May of 2008 and subsequently posed at least two suggestions to the developers about the problem and solution, as well as a few direct emails and countless attempts in_chat.

I still use bands on ocassion - for the obvious sheet hold operations, etc.), but in general not in the way I'd like to have been able to, primarily because of this lacking. It makes little economic (in all senses of the word economic) sense to spend hours being frustrated by a snap-to function that serves to do little more than act as a psychological (and physical) barrier to producing a result, especially when the game was promoted heavily as a _direct_ human-computer interactive, rather than a continuance/variant of Folding@Home. I'd rather just tweak side chains, bash the protein a bit, and get back to more productive uses of my time than to sheep into a folding at home position due to (imo) purposefully, poorly engineered HMI and the "group" model.

What's more, the solution to the snap-to secondary issue with bands (I *was talking about a solution to the glaring issue in the rotate and optimize, ie, straighten functions) is quite trivial to implement, since the program is obviously using a variant on the Gravity_Well based "snap-to" function for ensuring adhesion to residue elements. Either one of two solutions could be very easily implemented and used to reduce this tedium

1) a toggleable release. Preferred because it is more ergonomic and reduces the likelihood of prolonged exposure carpals when a large number of bands are being placed with one end in open space.

2) a ctrl-key + mouse event to individually disable snap_to. This, however, has a subtle physical annoyance that can accelerate development of carpal tunnel syndrome for repeated use which should be obvious to anyone who uses a computer ritually...

The sole, almost trivial complication would be ensuring that the initiation point would always have snap_to, so that residues would be locked into the band at the terminus of the initiation end. Again, however, I believe that to be a trivial problem for degreed-program, university level or better programmers, and one that should be apriori _tantamount_ for those involved in the informatics department that is tied directly into this venture. In other words, for anyone and everyone involved directly in the creation and maintenance of this research program or the game itself.

Ideally, both methods would be present, #1 for the convenience and safety of those volunteer participants who care about their long-term physical (dexteral to be specific) health, and #2 for "one-offs" and people who don't care about their health or time, and imo should have been for the duration. It has been in CAD programs for two decades.

On a side note: As one who does periodically suffer from carpal tunnel syndrome as a result of over twenty years of CAD and general computer use as both an amateur (gamer, internet, computer art, computer music composition, etc.), and professionally as programmer, engineer, and now web dev, I can attest that the ever present pain in your wrists, hands, and arms is nothing to scoff at, and once you are "part of the crowd", it's pretty much too late to go back and will only get worse as you step over the line from youth tho true adulthood and then to "aging"...so anyone reading this, you've been warned.

fwiw, in case the thought occurs. I have already posted suggestions on more than one occasion and am highly unlikely to do so again. This is just history, discussion, and a few minutes of time-to-kill while I drink my morning coffee and wait for my support-level time period to roll around so I can address a problem with the hosting company that we use.

Ah...speaking of which, it's 8:58am. time to freshen the coffee, power up Skype and IRC, and dial the telephone.

Joined: 09/18/2009
Groups: SETI.Germany

We need an option to disable the auto-connection, when applying bands with the mouse.


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