restore best with conditions met

Case number:845813-986696
Topic:Game: Tools
Opened by:Judecca
Status:Closed
Type:Suggestion
Opened on:Tuesday, August 18, 2009 - 00:53
Last modified:Wednesday, February 8, 2012 - 23:16

On proteins where conditions must be met to get a score, it would be useful to restore the best solution where the conditions were met. I've noticed that the highest scores I get on the BETA Conditions puzzle don't meet the conditions, and hitting restore best often causes me to lose my best solution that did met them.

(Tue, 08/18/2009 - 00:53  |  16 comments)


Joined: 11/10/2007
Groups: Window Group
Topic: Developer Preview » Game: Tools

Changing project.

Joined: 04/13/2009
Groups: Contenders

Yeah -- I've lost a couple of my top solutions this way. It's annoying.

Joined: 11/10/2007
Groups: Window Group

Would you prefer another button, or to just replace the existing restore very best with this behavior?

steveB's picture
User offline. Last seen 1 year 10 weeks ago. Offline
Joined: 01/08/2009
Groups: Void Crushers

Hello Seth,

I would recommend a "set very best" new button to solve this problem, for the following reasons :

1) when the foldit algorithms are wiggling a puzzle with conditions not met(cnm), occasionaly the protein will distort into a shape where the conditions are met, just for a fraction of a second. If you are eagle eyed you see the red line flash on and off across the score. If the "set very best" was automatic then you could have the bizarre situation of starting with a structure at 9000pts(cnm), ending with a structure at 9500pts(cnm) but with a "restore very best" at an arbitrary point somewhere between the two.

2) the problem also applies to "quest to the native" puzzles. Often, whilst distorting the backbone of the protein to match the ghost image, the closeness of the protien to the ghost is much better than previously, but the score is significantly lower. It would be usefull at this point to "set best score" to what is displyed on the screen. The benefit of this is that currently the "set recent best" functionality of the programme is unavailable for use as a development tool if it is being used to store the best matching shape (which is the only way to use the programme at present).

Obviously if you implement a "set very best" button its going to need a warning sign the size of a small planet,

Rgds

SteveB

mat747's picture
User offline. Last seen 1 year 14 weeks ago. Offline
Joined: 10/15/2008
Groups: Void Crushers

I would prefer another button.

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

Another button,too.

mat747's picture
User offline. Last seen 1 year 14 weeks ago. Offline
Joined: 10/15/2008
Groups: Void Crushers

Still waiting

Joined: 11/10/2007
Groups: Window Group
Status: Open » Open

A fix for this will be coming up soon!

Judecca's picture
User offline. Last seen 1 year 13 weeks ago. Offline
Joined: 05/09/2008
Groups: None

May I also suggest a lua function for this new feature. If it isn't implemented now it will just be another feedback suggestion later. Might as well get it out of the way on the first shot. I'm sure the script writers will appreciate it

Tlaloc's picture
User offline. Last seen 2 weeks 3 days ago. Offline
Joined: 08/04/2008
Groups: Mojo Risin'

Yes...a button and a lua function, please!

Joined: 11/10/2007
Groups: Window Group
Status: Open » Closed

The newest update has a button and lua function for restoring the "credit best", the best solution that satisfies all the conditions and therefore gets credit.

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

Nice job, but we still need a "set very best" function.
An example:
If you do a rebuild, it often creates more than one result in the undo bar.
Most times, the first rebuild results are better than the last ones, but you can only get the very last rebuild result by script, as no other script commands are executed, when a tool is running.
Setting the "very best" state before rebuilding and getting the very best result at the end of rebuild would fish out the very best of all rebuild states which were generated by the rebuild command.

Kind regards,
Alex

spvincent's picture
User offline. Last seen 10 hours 24 min ago. Offline
Joined: 12/07/2007
Groups: Contenders
Status: Closed » Open

Thanks to the developers for addressing this issue, but the implemented solution, restore_credit_best(), simply doesn't work as far as scripting is concerned. The main issue is that the user will be forced, for many scripts, to start at the best solution rather a lower-scoring one of their own choosing, making exploration of alternate structures more difficult. There are other problems too.

Preferable would be a function such as is_condition_met() which would return a boolean value. Much more flexible and hopefully easy to implement.

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

Just as spvincent said, scratching at the top the whole time doesn't help.
We definitely need a possibility to change the "very best" state.

Tlaloc's picture
User offline. Last seen 2 weeks 3 days ago. Offline
Joined: 08/04/2008
Groups: Mojo Risin'

There needs to be two functions:

1) are_conditions_met()

2) get_credit_best_score()

While you are implementing that last one, can we get

3) get_recent_best_score()

4) get_absolute_best_score()

as well. These should all return negative scores or have the argument that allows negative scores.

beta_helix's picture
User offline. Last seen 18 min 2 sec ago. Offline
Joined: 05/09/2008
Groups: None
Status: Open » Closed

the requested functions should have gone out in Lua V2

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