A question about what it's actually DOING
entropy
Yah-Der-Hey (Wisconsin)
Ok, I think I'm going to at least poke around with folding again. But I gotta know something that's bugged me for awhile beforehand.
F@H simulates proteins folding. Ok. So. What's that mean? What *exactly* is it doing? How can it be beneficial at all? If we can simulate it, we must know how it works...
If we're looking to see how diseases (certain) diseases are formed, apparently it's from proteins 'misfolding'. So, the only way I can see us simulating that is to simulate it in a 'live' environment and tell it to go to work, while introducing different variables (heat, movement, acids, water, whatever) to see if it will screw up...
So, before I start this again, I need to know this. What is really happening, and how's it beneficial? (every site says the same thing - "simulated protein folding." Lovely. Now what's that mean?
F@H simulates proteins folding. Ok. So. What's that mean? What *exactly* is it doing? How can it be beneficial at all? If we can simulate it, we must know how it works...
If we're looking to see how diseases (certain) diseases are formed, apparently it's from proteins 'misfolding'. So, the only way I can see us simulating that is to simulate it in a 'live' environment and tell it to go to work, while introducing different variables (heat, movement, acids, water, whatever) to see if it will screw up...
So, before I start this again, I need to know this. What is really happening, and how's it beneficial? (every site says the same thing - "simulated protein folding." Lovely. Now what's that mean?
0
Comments
This process of protein folding is NOT understood completely. It happens SOO fast, think about it, a freakin nanosecond for petes sake, that its really hard to gather data about the whole process. Gathering data at a very low level like Stanford is doing now was not available in years past. The tools and computing power just wasnt there for them to do that kind of thing.
For a good chunk of FAH's beginning all the data we were sending back to Stanford was actually just to test to see if what Stanford was simulating is actually what happens in the human body. And what they have been simulating is IN Fact the same in real life. Remember proteins and the human bodys is REALLY complex. There are soo many variables, most of which we cannot even think of. But somehow stanford has been able to reproduce a sample protein simulation. It took them several versions of the client and over a year just to see if what they have been doing is actually happening inside our bodies. Thats how complex proteins are.
Now the whole process of protein folding is not completely understood. If it were and we know they misfolding causes this and this disease, wouldnt you think they we would be able to find a way for this misfolding to stop? But we dont even know how the proteins are misfolding in the first place. Thats the first step. Knowledge has to come first. Once we understand why things happen, then we can go about the business to fix the problems. We cant fix something that we dont know what exactly is wrong. Its like a car. It wont start that may be obvious like the diseases but the reason behind it may still be clueless the the person with the dead car.
With enough simulations, enough as in like millions, hopefully the guys at Stanford will be able to understand this very importand and very complex molecule. The more simulations the better. Knowledge is power, power to change things. Power to fix things.
If you want to know what we are actually calculating I believe we are calculating the movement of the atoms in the protein molecle. The movement is in 3d so that would mean vectors. And the vectors are not going to be whoel numbers but rather real numbers with decimal points. Thusly floating point numbers. AMDs have the better FPU compare to Intels. That is why AMDs are better folders per clock cycle since so much of the calculations are floating point numbers.
Hopefully that all made sense. Its not like 4:37 in the morning and I was up partying last night and I was also talkin to Geeky on MSN in the middle of all this. Time for bed.:)
That's some multi tasking you've got going there mmonnin.
I 'kind' of thought it was something like that but your explanation clears it up for me. Thanks.
.................."If you ain't folding, you ain't no friend of mine".
What do we start with, what do we end up with, and what happens in between (from the time we get a WU to the time we finish it)?
I guess I just don't completely understand what it is we're simulating, or calculating, or whatever. And from how I'm thinking of it right now (which, granted, is probably wrong) it doesn't seem like it could be of any use at all
(Oh, and it was "We can fold if we want to. We can leave your friends behind. 'Cuz your friends don't fold, and if they don't fold, well they're no friends of mine" I believe
Entropy, essentially what each project does is a series of tests to see how to prevent misfolding. and of a certain protein chain that does misfold and is known to do so in some cases, or different ones that do so in a set of projects.
Can changing saline content in water, indusing local high heat, etc, stop certain misfolds that cancer CAUSES TO happen???
Well, scientists could take weeksand months and years of experiments to find out, or the over 165,000 ACTIVE folding boxes (at any given time, AT LEAST that is the average number folding right now in teh distributed network). What are our boxes doing??? Testing scinetitific hypotheses for medical research institutions. Essentially they are doing what science would do, altering one variable at a time based on previous results looking for things that when combined do stop folding misfolds of one kind-- each project set can in fact be for a different kind than another, or a continuation of a project where certain things proved to work better, to see how they can be combined with others.
How does this help??? Well, to do it for real, either animals or humans would have to be used. Second, the boxes can compress time that scientists would have to wait for real-time experiments to produce results.
That is not to say that real experiments will not follow, indeed they have and will-- based on lnowing what sets of things work more in simulation before running experimental series that in fact could take decades to produce results if run in a sequence string in real time.
So, how do you get checks and balances in this??? at any given time, many more than one box are running the same basic sequence of sim. We do not depend on the results from a single box, nor from just two, nor from just ten. And, WE DO NOT WANT TO-- isolated bed results can give false signals to a whole direction of research, but if 20-30 boxes all come up with a null experiment, we know what NOT to work on, and if they DOP come up with a way to keep a certain biochemc folding sequence of misfolding kind from happeniong, then more projects exploring that are launched.
EXACTLY what are we doing, we who fold??? Vijay Pande is an assistant prof at Stanford, specializing in Chemistry, Bio, and Physics. He's working with team of folks, some of them grad students, some researchers with other institutions, some other professors. He and his group is sponsored by Stanford to a degree, they provide some funds for equipment, some infrastructure, a building to house the stuff in, some funds to teach students who want to specialize in computer modelling research of medial or biochem kinds, and some of the projects we fold can have grad student input or input from pro researchers, etc.
Our boxes are following the lead of Vijay and his team, doing dynamic reasearch, in some cases determining the fine facets of how to interrupt the misfolding of proteins. Dynamic research in fact is intended to NOT echo totally preknown things, instead, to use what is known and add something and see what happens. Thus, from day to day what folding does differs in fine detail.
As I said, I am not Vijay, and in fact I am a folder myself and not someone with a doctorate in anything.... I've got thirty plus years looking at and building and playing with the system aspects of computing, though, and can safely say we are narrowing the field of knowledge that needs to be researched in real time at minimum, and at best a few fewly major things have been found that helped make treatment better for folks because real LIVE experiments confirmed the folding results. Who would experiment on a human??? A doctor trying to save that human who had no known canned and proven treatrment for what he or she faced-- and might do it with the patient's (and the patient's relatives) knowledge and legal consent, in fact a lot of cancer treatment goes like this-- the larger rules are known, but the fine details need to be determined. So, if boxes can do it, real folks get not exposed to the things best not done that have been ruled out by our boxes simulating hypotheses within the framework of what has been proven to be theory that is situationally valid in real life.
Just as some motor vehicle companies do air tunnel sims of vehicle body shapes, and aerospace engineers prevet airplane and rocket and shuttle shapes with computer simulations before even building major sized models and running them in REAL air tunnels and then taking the bad aspects and taking them out of the design before the design finalizes into reality, so in a sense are we helping remove some of the most dangerous to human treatments with the chance of finding something uniquely good and able to do what the whole distriubted network of boxes is trying to do-- get treatments that kill JUST cancer, JUST Altzheimers, JUST Avian Flu, JUST the disease being treated without wearing down the whole body so the patient expires in mid-treatment.
Is this kind of sim experimentation dangerous??? No, because it is in sim and results are being confirmed before being used. IF folks were going from one sim to a real experiment, that would be dangerous, VERY, but CANCER is very dangerous also.
Do some projects do more than just determine direction???? YES. Let's say a bright research team findas something that does work in a limited number of patients. Can thta then be simmed and possibly in situations thta vary slightly to see how likely it is really work for many folks??? YES. And that is in fact part of the jobs that boxes that fold do in some projects where research institutions sponsor the work on folding thus giving Vijay's Team more funds for more servers, etc.
Can I tell you exactly what p1306 is??? No, but I can tell you that p1124 is subtype 4 of a continuation of project 112.... And that the whole 1300 series is based on other work and continues it. I can only give you vingettes or glimpses into the categories of things that folding does, I do not have a doctorate and am not the Einstein of BioChemPhys, and even Einstein knew of uncertainty.... And some uncertainty elimination is what our boxes are doing at worst and best they help guide real researchers or confirm and in some cases delineate the edges of the scope or limits to what should be a good treatment given what IS known.
With different simulations they can add different variables into the whole process. I know for one set of WUs we were doing for awhile had water in the title so they must have been experimenting with that.
Think of it kinda like a game. Each atom is given rules for what it can do in certain situation. When this and that variable are equal to a certain value each different atom can perform a certain movement. If one of the variables is different it will take a different path. As in a game, the position of your character is one variable and the monsters will most likely take the path towards you. The game physics determine what happens. Same thing for the Atoms in folding. Stanford gives rules that are just like the rules of our body. They will follow those rules, because well...they are rules.
We are most likely get a molecule in a certain position and it will proceed, following the given rules, for so long until the Work Unit is done. The FAH program is basically a big loop. Here is an example of the end of a Work unit.
[16:06:51] Completed 196000 out of 200000 steps (98)
[16:28:29] Writing local files
[16:28:29] Completed 198000 out of 200000 steps (99)
[16:50:17] Writing local files
[16:50:17] Completed 200000 out of 200000 steps (100)
[16:50:17] Writing final coordinates.
[16:50:20] Past main M.D. loop
The whole program runs the big loop 100 times and then it knows when to quit. In the 2 bolded lines it says its writing the final coordinates. Coordinates of what? Most likely of the hundreds maybe thousands or maybe even millions of Atoms in a protein molecule. (I am not sure if you realize the size of Proteins. They are some of the biggest in the body and very complex.) Once its done with the loop it has simulated a certain time frame of the folding process and it sends back to Stanford where everything ended up.
Now as I said before in my other post, they for a long time we were just basically testing to see if our simulations done by all of the users is correct and is comparable to real life. Stanford has written it so that they do match and are able to take our results and find out where the atoms move during the whole process. They can add/change different variables to see what happens.
How does this help Stanford? Well look at all the thousands of computers running the FAH project. Think of the computing power each one of our PCs has and how long it takes for each of our high end machines to complete the process. And what we do in one Wu is NOT the complete process of one protein folding. We here at Team Short-Media have completed 625k Work Units. Thats a lot of Work Units. Add the top 100 teams together and you are looking at 100 million or so. Thats AN ENORMOUS amount of computing power.
Now if Stanford didnt have the thousands of PCs working on the project, how many WUs do you think they could have done by now. Go look up "Pande Lab." They have completed 77k WUs. Thats nothing compared to what some teams do. The top teams do what they have done so far in 2 whole weeks.
Stanford could not do this on their own. We provide Stanford with the computing power to continue with their research. WE fuel Stanford's research. Without the FAH project they wouldnt have enough data and it would be pretty much uesless to continue their research. That or they would have to find another way to gather data. They sure couldnt run simulations on computers.
I hope that clears some more things up for you and gives a huge reason to continue folding and to help Stanford out. They NEED us!!!