4 replies [Last post]
brow42's picture
User offline. Last seen 23 hours 54 min ago. Offline
Joined: 09/19/2011
Groups: None

Hi gang,
I had a short chat with David Baker and grad student Javier Castellanos this morning. We discussed new visualizatoins options for foldit, design aids, and ways to facilitate more communication between the community and techical staff.

Design Puzzles

David once again commended players for their interesting designs and how much they have improved under the filters, and hopes they will become much less painful soon as the filter effects are moved into the score function itself. I didn't have any specific comment about the filters right now, but I'm sure many of you have plenty to say!

I had a suggestion for the recent sepsis puzzle (and other ligand bonding puzzles). I noted that a good solution seemed to require 1) the hydrophobic core remain covered, 2) we need increased bonding to the ligand, and 3) we need a nice stiff secondary structure to hold the bonding residues in place. We need more residues available to do all three (the last two puzzles were +10, then +6). Also, I've noticed that the last locked residue would leave the protein pointing in an 'inconvenient' direction, and more should be unlocked farther back.

Design Rules

The Baker lab recently did a study on how groups of strands and helices folded, and made a list of rules summarizing them. They then designed entirely artificial combinations following these rules, ran them through Rosetta@Home, and then actually synthesized them in the lab and measured their structure, and found that the designed proteins folded as designed.

These rules answer simple questions: should this strand turn left-right; should this helix go over-under the sheet; should the helix go left-right. The answer is always "both" but short connecting loops prefer one way, then prefer the other as they get longer.

Since the Baker lab wants us to not just make interesting designs, but designs that fold as predicted, I asked if these rules could be turned into simple English, along with visualization aids (more below), so that players could make more stable designs. Javier, who does design work, could help with this.
These rules would also make de novo folding less daunting. Perhaps, if the visualization was kept informative, not suggestive, it would preserve diversity.

Visualization

I asked David if we could have Foldit provide more information to the player. For example, when I started as a new player, I would see that a segment has a bad backbone score, but I wouldn't know why. Eventually I learned that if you left a script to rebuild that segment, eventually it would get fixed. There are many sorts of twists the backbone can make that are invisible. Foldit could communicate more information about the angles, and highlight flipped segments (cis bonds).

I emphasized that whatever visualization is added, it should not be color based. Now that I think if it, it should not add clutter, either. Perhaps it draws the extra visualization only when a key is pressed.

One very easy visualization to add is to simply indicate the direction of the strands (which are drawn as arrows in pyMol but are unmarked in Foldit).

Community Interaction

David Baker would like this chat to be the revival of the player-dev-scientist chat. ( http://fold.it/portal/node/992501 ) He would like the community to participate in the bi-weekly skype meetings. The group leaders and expert players could take the comments of the community, consolidate them, and translate them (for example, turning what a new players sees happening into what foldit is actually doing). The scientists can't really monitor the forums. Bug reports are immediately looked at by the devs, but questions and suggestions aren't necessarily passed along.

I think a good start would be: the expert players consolidated the questions and suggestions (not bugs) in the feedback, the group leaders contribute what was being said in group chat, and the community in general use the existing voting mechanism on the questions and suggestions. This could be presented as a consolidated posting or in Skype. I prefer written communication, but David likes Skype, and with Skype you know they've gotten our feedback and discussed it.

Please post your thoughts on how we can organize ourselves to communicate more directly and regularly. Also, what visualizations you'd like to see, or if you think design rules is a horrible idea, post below!

-- Brow42

marie_s's picture
User offline. Last seen 1 year 42 weeks ago. Offline
Joined: 05/18/2008
Groups: None
Skype

On community interaction,
As you know, Foldit community is from many countries.
Many player, like myself, doesn't speak english very well.
Skype is not a good way to communicate for us.

Design rule and vizualisation
I do not think that teaching that here is good way to fold and a not good way to fold is a good idea.
There are 32 tutorials, they could be better but after that, let people play and learn by yourself.
If they want, they can join senior players on tam, read the wiki,run recipes... but I think that the more player learn by themselves, the best are the luck that they choose new way to think and find solution to problems biologists cannot solve.
If I had to learn a set of "rules" or have advice like "run recipes, they solve problems for you", I will not have play for a long time.

Joined: 09/27/2015
Groups: None
Skype

I think Skype is a good way to communicate. You can have a Skype conversation in English and get it translated to another language real-time via the Translator feature. I found out about this during one of my classes at Preply.com . They recommended it for practice. Skype tries its best to bridge gaps.

Angus's picture
User offline. Last seen 12 hours 58 min ago. Offline
Joined: 06/04/2008
Groups: Beta Folders
Skype, and meetings in general

I think Skype would be my distant third choice for conversations with UW. I would prefer to see something like WebEx or Lync that includes VOIP and/or toll-free phone connections. Seeing things demonstrated on a shared desktop and being able to ask and answer questions is a much better way to collaborate than just voice, or text via IRC which become confusing very quickly with multiple threads going at once.

Whatever rules or visualization improvements are made, they should be in very simple non-scientific language. Even reading brow42's message above is at or above the limit of a normal players comprehension, he's talking about things that I can't figure out. I'm not an "expert" player, but I've been here since the first few weeks, and if I have time to play I can stay reasonably well placed in the standings. But, I have done that without any scientific knowledge of what's happening. I couldn't name off the amino acids, need the mutate wheel to tell you if they are hydrophobic or hydrophilic. Talk of "angles", "cis bonds", and whether a helix should go left-right or right-left because of some unknown context simply are babble. I have to think that is more representative of most players, newbies or experienced. Those who really understand the science are a rare few, and the rest of us will never be able to compete with them or understand what they are talking about. This is a hobby that I'm addicted to, but I don't intend to try to get the equivalent of a degree in computational biology just to stay "in the game".

Oh, and so far, in 5 years, I have *never* been able to participate live in a conversation with the devs or scientists, they are always during regular work hours. Can we at least move them around so they are in the evening for the US audience every so often? 6PM PDT is only 9 PM EDT, so that gets the entire home market able to participate. I've toyed with the idea of taking a day off and driving the few miles over to to the campus, but feel I would just be wasting their time since I can't talk in scientific terms.

brow42's picture
User offline. Last seen 23 hours 54 min ago. Offline
Joined: 09/19/2011
Groups: None
Info Presentation

I don't disagree with Angus. One of the truly, brilliant things that Center for Game Science did, and why they deserve all the press they got, was that they turned a tremendously specialized field with horrific math into "Shake & Wiggle", and got a simimian immunovirus structure solved with nothing but that.

My complaint is that the client collapses lots of problems into a single number "score" which has a color "red". I would like to know what to do with that. It doesn't have to be technical. It SHOULD NOT be technical. They turned a horribly complicated math problem into "shake and wiggle", I'm sure they can highlight badness like cis-bonds in a similar way. (except sometimes cis-bonds are good, which is why I brought this up with the scientists and not just the devs & players)

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