Placeholder image of a protein
Icon representing a puzzle

655: Blind Electron Density 2

Closed since over 13 years ago

Intermediate Overall Prediction Electron Density

Summary

Created
November 20, 2012
Expires
Max points
100
Description

This is our second electron density puzzle with real data. This should be easier than our last density case, since the scientists were able to pinpoint which regions belonged to a single copy. Try to fit the protein into the density to get the best score!

Top groups

  1. Avatar for Void Crushers
    1. Void Crushers
    100 pts. 10,938
  2. Avatar for Anthropic Dreams 2. Anthropic Dreams 81 pts. 10,915
  3. Avatar for L'Alliance Francophone 3. L'Alliance Francophone 64 pts. 10,442
  4. Avatar for Gargleblasters 4. Gargleblasters 50 pts. 8,293
  5. Avatar for Go Science 5. Go Science 39 pts. 8,267
  6. Avatar for Beta Folders 6. Beta Folders 30 pts. 8,247
  7. Avatar for SETI.Germany 7. SETI.Germany 23 pts. 8,066
  8. Avatar for Contenders 8. Contenders 17 pts. 8,051
  9. Avatar for foldeRNA 9. foldeRNA 12 pts. 7,562
  10. Avatar for Deleted group 10. Deleted group pts. 6,730
  1. Avatar for steveB
    1. steveB Lv 1
    100 pts. 10,936
  2. Avatar for utaca 2. utaca Lv 1 99 pts. 10,935
  3. Avatar for karstenw 3. karstenw Lv 1 97 pts. 10,791
  4. Avatar for Timo van der Laan 4. Timo van der Laan Lv 1 95 pts. 10,758
  5. Avatar for Susume 5. Susume Lv 1 94 pts. 10,567
  6. Avatar for Madde 6. Madde Lv 1 92 pts. 10,552
  7. Avatar for Ch Garnier 7. Ch Garnier Lv 1 91 pts. 10,437
  8. Avatar for gmn 8. gmn Lv 1 89 pts. 10,381
  9. Avatar for MooMooMan 9. MooMooMan Lv 1 88 pts. 10,281
  10. Avatar for brow42 10. brow42 Lv 1 86 pts. 9,299

Comments

brow42 Lv 1

I really liked this puzzle. It had exactly the right amount of detail I need. You could see the backbone and some sidechains. I'm not anywhere near the high score but I think whomever has the high score will have good results for you.

The last blind ED puzzle (on a lattice) was too complicated for me to play, it was just random blobs. Also, the simulated noisy ED in the past were also too noisy to play. But this, despite being noisy real data, is quite playable. I could see sheets and helices clearly in the ED. I'm pretty confident I'm within 1 segment width for my puzzle.

However, I couldn't have done it without Susume's jpred prediction. There are so many ambiguities in denovo folding, starting with "which end is first". I'm sure only the most experienced folders can start without a structure prediction. Sure, it was mostly loops = no preferred structure, but knowing where the helices and probable u-turns were, was critical.

Have you considered pre-smoothing the ED? That might be important for taking advantage of human abilities.

Thanks again for a fun puzzle of real data!

P.S. I was able to do this puzzle. But maybe the puzzle was easy? Can you say if computers were able to fit this ED with no trouble?

alwen Lv 1

I agree with Brow, this one was fun especially once I saw SteveB's screen shot early on and started to get a feel for hauling the denovo into the cloud.

(I've entered a feedback about the Align to Density button, though. Compared to the alignment tool, this button does not align the protein to the density, despite the text on the button.)