16 replies [Last post]
Joined: 04/24/2014
Groups: None

This thread is specifically for Susume's Electron Density Video. Thank you Susume!

- All general questions/suggestions/concerns/commentary should go to the puzzle thread. We also recommend you bring your general questions about electron density here to our science chat on Friday.
- All bugs should either be filed in the puzzle thread or feedback

Let's hear your questions about the video!


Angus's picture
User offline. Last seen 5 days 7 hours ago. Offline
Joined: 06/04/2008
Groups: Beta Folders
I still have no idea how

I still have no idea how random bits of blobby things in the bigger wooly blob can be identified as sheets. I think that whole process needs a LOT more explanation.

alwen's picture
User offline. Last seen 6 weeks 6 days ago. Offline
Joined: 10/03/2011
This is really hard to

This is really hard to express without an electron density cloud to refer to, but cloud sheets, just like in Cartoon view, will zig zag back and forth. Sidechains usually show as lumps on opposite sides.

This cloud has good resolution, so it is possible to identify some of the paddle (phenylalanine/tyrosine/tryptophan) sidechains.

If you look at a protein in Sphere view with sidechains showing, that is the one that most closely matches the cloud view, big and lumpy.

Susume's picture
User offline. Last seen 3 hours 8 min ago. Offline
Joined: 10/02/2011

I don't know any way to get around the need for practice. Using the tools to narrow the field of view to a slice makes it possible to see shapes, but that by itself doesn't make them easy to identify. Hopefully seeing some samples and hearing some descriptions of the kinds of shapes to look for will help. Like any visual pattern, the first few are the hardest to see, and the more of them you see the more visible they will become.

The second video (on sidechains) also offers some views of a single sheet, with sidechains hanging off on either side. Maybe that view will be easier to recognize.

Joined: 09/24/2012
Groups: Go Science

Now I'm impatiently waiting for the next video subsume introduced on the end of this one ...

My main question would be for mid game: Now that I have many helices and loops in the cloud, how could I find the place for the sheets, since my all protein is interfering in my available views (slices etc). The method explained by subsume is good for start (even if you can misinterpret the big structures and start in a completely wrong direction - as I did :). How to use it at mid game?

Joined: 09/24/2012
Groups: Go Science
related viewing suggestions here:
Joined: 03/30/2013
Groups: Go Science
Totally want to say THANKS

Totally want to say THANKS Susume. I have been playing for about 2 years and I didn't even know you could add dots. Also I have other experiences navigating electron density in 'O' and COOT, which probably makes me different than most foldit players. I have been finding it frustrating to navigate density in foldit AND THIS VIDEO HELPED ALOT!
I want to beg Bitspawn to also do a video next! Is that possible? I have watched Bits dominate these kind of puzzles for a while now and would like some insight into the secret sauce.

Joined: 08/09/2010

Could someone transcribe the video?
I do not understand a word O:-) and with the text I would like to put subtitles to this video.

Thanks :)

Susume's picture
User offline. Last seen 3 hours 8 min ago. Offline
Joined: 10/02/2011

I have a transcript now for the first video, but I can't attach a file here in the forum. I will try to find a way to post it.

Joined: 04/24/2014
Groups: None
Here we are!
Joined: 09/24/2012
Groups: Go Science
there is a tool

I had an option on the video viewer that automatically added a subtitle (I had no sound on this computer).

The English is quite aproximative but it was almost enough for me to understand what Susume explained.

There is even an option for translation but I did not try it (it would be worse than this automatic transcript).

Joined: 09/24/2012
Groups: Go Science
The secret of pushing the segments inside of cloud parts

Now that I could recognize some structures and sidechains, my problem is to put the segments inside of the density.

1-do you push and pull it manually after freezing the extremities? (it does not work well! not precise enough, and the all protein is moving)

2-Do you cut it in pieces and push it inside before to uncut? (it does not work well !)

3-CI on 0?

4-Do you freeze extremities then add a lot of bands in all directions, then wiggle? or rebuild? (it might work but it's very difficult since the front and back are not visible and put the segments in the blind)

5-Even with this in mind, it often happens that the structure length doesn't match. For example, my helix is longer or shorter than the related density. After many attempts, I don't succeed tu pout it "exactly" inside.

6-Big problem with the sidechains ! We have a backbone and we recognise a succession of sidechains in the cloud. But the position of the sidechain itself varies a lot. It's really not easy to put the backbone on the right place with sidechains going in wrong directions.

Thanks again for all your help ! Current results suggest that only 3-4 players (with > 17000 pts) are able to hand fold correctly on density puzzles ! That's not enough !

Susume's picture
User offline. Last seen 3 hours 8 min ago. Offline
Joined: 10/02/2011
Cut points

I use many cutpoints (maybe 40-50 for a 200 segment protein) to let me match the cloud with small pieces of backbone. If a small piece of backbone will not allow matching the sidechain positions, I will band those sidechains to their place in the cloud and rebuild (by hand) only that small piece of backbone until they match better (disable cut bands while rebuilding). Once I have matched as much as I can see, I use zero-length bands script to band all segments in place (band strength 10), set CI low (0.2 or lower), and shake/wiggle until score stabilizes. Then close cutpoints, wiggle some more, weaken bands, use DRW script on red areas. Then I run DRW for a long while (still at ci 0.2) with long sections (7, 8 or 9) to let it settle things better in the cloud. Then I look over the whole protein for places where I can improve the fit to the cloud. Then back and forth between scripts and hand-fitting until I am happy with the fit.

Joined: 09/24/2012
Groups: Go Science
Wow !

I saw the second video. It's like "lesson 2". Impressing technique ! Thanks also for the full explanation on cut points above.

If I find time, i'll try to "learn by doing" like you in the videos for this puzzle.

I think that this is the best way to learn, using the same puzzle and your video.

Looking forward for lesson 3 !

Joined: 04/24/2014
Groups: None
Joined: 04/24/2014
Groups: None

If you're watching the videos and need captioning, we're matching the transcript to our videos (because autocaptioning is so very terrible), and if you're interested in translating this to other languages, please send me a private message!

spmm's picture
User offline. Last seen 8 weeks 3 days ago. Offline
Joined: 08/05/2010
Groups: Void Crushers
final score influences

Curious to understand the contribution from the group AD to the final score that Susume has reached.


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, Boehringer Ingelheim, RosettaCommons