Several miscellaneous folding questions

[fgiouh89fgh]

1# Is more focus being spent on Alzheimer? Seems many of the machines I got folding are doing Alzheimer recently. Is this a sign perhaps we are getting closer to figuring out Alzheimer proteins we are folding?

2# Why are some atoms in the "animated viewed" very small? Seems there are commonly oxygen, nitrogen and carbon atoms. There are also whitish-gray (hydrogen) atoms, but there are also these tiny whitish-gray atoms. Why are they so small sometimes?

3# How can simulate protein folding? What are the variables that change that can make atoms fold differently that require simulations?

4# How do we know our simulations are representing misfolded proteins and its not something like cosmic rays, radiation, or something else causing misfolding?

5# I see this in my logs: (Run 15, Clone 61, Gen 99). What do they mean? Is max WU's needed to solve something: maxRun * maxClone * MaxGen * numberOfSimilarProjects?

6# If we are simulating something to figure out how it works, how does folding not go on forever till it cant fold, instead of 0% to 100%?

7# Sometimes project dont hit 100% because proteins cant naturally fold any farther. But how do we know its not just a random inaccuracy in a CPU/memory of the contributor?

[Zagen30]

#1-3 I'll leave to someone more knowledgeable.

#4: Some (many?) of the results have been verified by comparing them to known non-simulated results to make sure the simulation methods work properly. Presumably enough results were verified to give the Pande Group confidence that their methods are correct. Every time a new core or method is implemented, the first set of projects using it typically rerun an old simulation to make sure they produce accurate results.

#5: The wiki explains it best.

#6: I'm not quite sure what you're asking. It sounds like you're thinking that each WU is a full simulation of a folding process, which it's not. As the Run/Clone/Gen wiki explains, each WU is a tiny timeslice of an overall trajectory. While many Early Unit Ends are caused by unstable hardware, some of them are simply the trajectory reaching the end of its process by entering an illegal state.

#7: There are checks in the software that detect errors in the simulation process. There may be other stuff I'm not aware of.

[bruce]

Regarding #1
The Pande Group is made up of a number of researchers. Some study Alzheimer's; some study other things. Also, research goes through many phases. First there's an interest in answering a specific question. Then there's a period of study what other researchers have already determined. Then the question is formulated into a plan and various FAH projects constructed. Those projects are distributed and results begin coming in. After a while, the question might be answered but more than likely, new questions will have to be asked. If conclusions can be drawn that represent an advancement in science, a paper is prepared for publication in a scientific journal. The paper is distributed for peer-review and most likely revisions will need to be made. Eventually the findings are published and the process starts over.

A lot of time goes into the initial formulation of the theory and also into the analysis of the data and the writing of the paper. We only see that portion of the time when the proteins are actually being simulated.

I'll try #3:
If you climb to the top of a nearby mountain with something like a huge beach ball and throw it down a steep incline, it will bounce many times as it descends the hill and it will eventually stop. If you do it again, there will be random variations in its path and it won't end up at the same place. If you repeat that process many, many times, you can determine that there is a statistical pattern to the final resting points and there will be a statistical scattering in the amount of time it takes for the ball to stop.

If there is a deep depression near the stopping point, most of the time the ball will fall in this hole but occasionally, it may end up in a nearby depression and never reach the lowest possible final stopping point.

There is a significant statistical nature to atomic motions much like the motion of the bouncing ball on a slope in my example. Normal folding is like all the times that the ball ends up at the lowest point. Misfolding is like when the ball ends up somewhere else.

I''ll also take a shot at #6/#7:
Computers do make errors and that may lead to a simulation that cannot reach 100%. Simulations which do not reach 100% are assigned to someone else to see if the EUE was caused by a flaw in your hardware. If everyone finds the WU cannot be completed, then there is something that can be learned from that particular WU. If other people can fold it successfully, it is discarded, assuming it was a problem with your hardware.

EDIT: Rather than thinking of a beach-ball coming down a mountain, it might be easier to think of the Mars Lander that was wrapped in Airbags and dropped by NASA from a parachute. It bounced for quite a while before finding a place to stop, and logically, it might have had a higher probability of stopping in the bottom of a depression than at some higher point, but the beachball is closer to what we normally might experience. In either case, the concept of mountains and valleys is a reasonable explanation of a 2D projection of the energy landscape that a protein encounters.

[Zagen30]

For #3, the wiki article linked above mentions some of the factors that can affect folding, such as temperature, initial velocities of the atoms, and mutations in the protein structure.