This is the post you'll want to bookmark and share with your fellow community members who may breeze past the news section in a rush to get folding!
The Date: 16 October 2014 (Thursday)
The location: #veteran, IRC (Get help with chat here.)
The Time: 1300-1400 (or so, but the official chat ends after an hour) Pacific
The Time Zone Converter: Right this way!
The Topics: Determined by you. Add your questions to this thread! As we're looping in our Foldit scientists specifically for this chat, they want to know what's on your mind. Looking for inspiration? Follow-up on previous topics? Read our previous chats.
Speaking of chat and veteran status - we've looked at how it was awarded, and made a few adjustments to ensure once you have it, that it shouldn't disappear! (This was a thing that affected some of our users.)
Let's keep the downtime issues and suggestions in that thread: http://fold.it/portal/node/1998425 - please note no scientist data was lost.
Thanks for the questions about the design testing, we'll see what information we can find out!
See: http://fold.it/portal/node/1998486
I would like to see his question treated in the dev chat or answered in a blog post.
1. Perhaps the scientists could incorporate a "fold" command that allows us to designate a "phobic" and a "philic" side, area, or region of a peptide, and the fold command could orient the chosen peptides appropriately according to that designation.
2. We need a more powerful LUA Math library - Sine/Cosine/Natural logarithm/Sigmoidal Curves (Synthetic memory), Fast Fourier Transforms, etc
3. Some players may not be good at programming but excel at mathematics. foldit should develop a tool to interface with them.
4. Maybe represent hydrophobic areas surrounding the residue as a colorized van der waals interaction radius / isosurface! Or even better, a common imaginary surface or plane / region shared by all 'phobics. Making it easier to "see" which way to best arrange the peptide.
5. Remove filters! Please integrate these into game play. It seems too much like a game *patch* . If these aspects are important - integrate them better into gameplay.
6. The scoring system should become a search-based model, not a computational one: Similar results to past solutions should hold more weight than numerical calculations based on compactness, hiding, etc (i.e perhaps a more powerful inference engine) I'll explain, in a moment.
7. Currently foldit does not scale well when syncing multiple clients on multiple systems under the same username. s/b more cluster and multi-core friendly please. (Better thread delegation)
8. I submitted a paper to the Clay Math Institute for the $1 Million math prize 4 weeks ago concerning a proof of P=NP. My hypothesis relies on the inherit "Elegance" of NP and it's ability to "project" it's complexity onto P, in the form of a Hologram - a nondeterministic Hologram in which all paths are known to P yet in P only 1 is taken due to sufficient information to eliminate the others paths. Hence the "Elegance" of NP :) Thereby proving that all questions in life become problems of "Search" not "Computation". I didn't mention it in my paper but I believe the existence of Google proves P=NP. Google does not "Precompute" any searches. But instead utilizes a rectilinear system of automated reasoning and keyword-attribute tagging in tandem with stochastic analysis of these keyword insertions. Google then "searches" again for these attributes (exclusively!) to "find" the corresponding solution pageset. No computation in the classical sense is done.
9. Like any quantum operation, any operation done in foldit should be fully reversible. Rosetta should be able to resolve a new pose into its prior conformation without the need for an undo operation, thus keeping in line with the 2nd Law of Thermodynamics - unless irreversible enthalpy has increased (in the closed system) due to the release of electromagnetic radiation. The notion of its irreversibility should also be the hint of our heading into a local minimum !!!
For any complex problem in Life we don't need to "compute" its solution space in order to be able to traverse it. We only need to properly ontologically "transform", "map", or "project" one of the many ones we already have. My proof of P=NP also proves that Turing Models are sufficient to traverse any given solution space in this search.
The Seattle Times article http://seattletimes.com/html/localnews/2024389152_folditebolaxml.html says "And the lab has also designed a protein that protected mice against a lethal dose of the influenza virus. Foldit players contributed to that research, which is still in the early stages." Which flu puzzle(s) helped in the research, how did they help, and whose solution(s) have been used? Is there any publication from this research yet?
Please post more Ebola puzzles where possible and sensible. Focus on experimenting with that instead of repeating old puzzles right now.
Susume has a good point in posting the progress on the flu research (thank you Susume !). I'd like more updates on that. Not hearing about it is less motivating.
Status of any OpenCL / CUDA developments of the core Rosetta functions.
We've recently had a few quest to the native puzzles which the veterans probably haven't done for a long time, I haven't. I don't understand why, if I've pulled the backbone and the sidechains to overlap the guide, that the protein twists apart into something completely different when I take off the bands. Isn't there supposed to be an energy funnel? Isn't the conformation represented by the guide supposed to be at the bottom of it?
There is an observation by some high ranked players that running recipes at lower CI in the beginning or even in middle game gives better results. Is this something the devs are aware of, and do they use this tactic in Rosetta? Can they provide an explanation why this works for them. (I never did this, just trying it out now and observing that it makes more sidechain bonds)
Do you have any CASP results you can share with us? Were we close to the native on any puzzles? Do you have any score distribution plots you can share with us? See http://fold.it/portal/node/988112 and http://fold.it/portal/node/1998460 for more details.
I have noticed that as Foldit runs, sometimes upload_tmp_UNIQ* files appear and then disappear in the directory where all.macro, scriptlog.default.xml, and log.txt reside. What are these upload_tmp_UNIQ* files for? Are they transferring the latest structures from our Foldit clients to the Foldit server?
I ask because I think these upload_tmp_UNIQ* files are related to Foldit client crashes that have been occurring lately, especially on puzzles with less residues. See http://fold.it/portal/node/996649 for details.
