3 replies [Last post]
Joined: 11/26/2012
Groups: None

Just wondering what are some good late game recipes? I'm struggling to get past a score of about 12000

LociOiling's picture
User offline. Last seen 2 hours 12 min ago. Offline
Joined: 12/27/2012
Groups: Beta Folders

...also known as Banded Worm Pairs Infinite Filter.

I seem to recall KarenCH started the Banded Worm Pairs recipes a few years back. BWP uses one or two bands to distort the protein slightly, then does local wiggle, covering varying numbers of adjacent segments at a time.

The BWP concept has spawned a number of variations. Bruno Kestemont created the "infinite filters" version. The "infinite" part means it repeats various "patterns" of different lengths until you cancel. (The bands are random each time.) The "filters" part means it disabled filters (conditions) during wiggle and some other spots to speed things up.

BWP IF 1.4.8 is a version I created a couple of months ago that may have slightly better performance. Formula350 was the chief instigator.

Local wiggle recipes are good at the end, especially if you give them some time to run, as KarenCH noted a while back. Often the games are small (thousandths of a point), but sometimes they can find a big gain, particularly with the BUNs filter.

Other recipes to consider are BandFuze and Acid Tweeker. Acid Tweeker makes small adjustments, while BandFuze is more aggressive, trying four different strategies in succession to find a better pose.

Joined: 11/26/2012
Groups: None
Thanks Loci!

Thanks Loci!

spvincent's picture
User offline. Last seen 6 hours 35 min ago. Offline
Joined: 12/07/2007
Groups: Contenders
Rather than looking at end

Rather than looking at end game scripts, it might be best to concentrate on getting a good initial design: joining a team can be helpful here. The Foldit team have said on more than one occasion (e.g. here: last minute of LabReport #8 https://www.youtube.com/watch?v=DfTazi4OjCU&t=184s ) that refining structures beyond a certain point adds little to the scientific value and that you'd be better off starting a new solution from scratch; perhaps sharing the previous solution with scientists if you think its promising.


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