How is DF different from Folding@Home??

[Eriol]

As I've been looking around for the "most worthy" cause to dedicate my idle cycles to, I feel like I've narrowed it down to DF and F@H. Unfortunately, I'm having a hard time figuring otu the difference between the two projects! How are they different? What are they doing differently? Why is one "better" than the other (like you people aren't biased :p )? Personally.... I like looking at DF more, but I'd still love to know the difference between the two actual projects :) Thanks for all your help.

[Paratima]

OK, me first. The scientific merits of DF are discussed at length in the Educational (http://www.free-dc.org/forum/forumdisplay.php?s=&forumid=32) branch of this forum, as well as all over the place in the rest of the forum! :D

I think you will find that that is perhaps why we like DF so much. We all get to discuss the merits and shortcomings of the science and programmatic aspects of the project. It's occasionally quite a free-for-all, although we do try to keep it polite. :eek: The ultimate Yea/Nay, of course, is in the hands of the folks who run the project, notably Brian the Fist, aka Howard. But here, unlike F@H, we get interaction and feedback on a pretty direct level.

This assumes that F@H hasn't changed since I stopped folding for them a year and a bit ago. But at that time, the management there just wasn't responsive. Howard and friends are VERY responsive and that's what makes the main difference.

Everyone else can chime in now! :smoking:

[Eriol]

haha, ok thanks :) and though it's off topic from the thread, I figured I'd ask this while I have people's attention.... I just installed the beta 6 and deleted the files suggested (filelist.txt, *.log.bz2, *.val.bz2) and now when I try to run DF it tells me that my handle is currently invalid and to try and re-enter it, and when I do so it tells me to please enter the handle that I got on the DF website )which it is) - any ideas?

[Paratima]

Hmmm. No. Haven't tried out that logic lately. Could be something has slipped up in the beta version.

It sometimes makes noises about your handle when it can't phone home. Check your network connection.

You can also try editing a handle.txt file. Put in your handle & nothing else. Should work. I'll try some tests in the morning. Anyone else got an idea on this?

[AMD_is_logical]

Originally posted by Eriol
haha, ok thanks :) and though it's off topic from the thread, I figured I'd ask this while I have people's attention.... I just installed the beta 6 and deleted the files suggested (filelist.txt, *.log.bz2, *.val.bz2) and now when I try to run DF it tells me that my handle is currently invalid and to try and re-enter it, and when I do so it tells me to please enter the handle that I got on the DF website )which it is) - any ideas? The beta server was forked off of the regular server some time ago. Any handle recieved from the regular server more recently than that will not work on the beta server. Try signing up on the beta server:
http://beta.distributedfolding.org/login.html
and using the handle you get when you do.

Later, when the beta becomes the new client, the beta server will be deleted. Then you will need to use the handle you got from the regular server.

[Brian the Fist]

To answer the question you are really asking, the difference (among other things) is that F@H attempts to fold small peptides (not big enough to be called proteins) by parallel molecular dynamics (MD). Normally MD cannot be done in parallel but they have devised a clever way to do so. Unfortunately anything big enough to resemble an actual protein is still computationally intractable with this approach. Nevertheless we can learn many things about the protein folding pathways, etc by studying how these smaller protein fragments fold.

We on the other hand, especialluy with the new beta, are looking at the folding pathways of full (albeit small) proteins. Our ultimate goal originally was to predict the final folded structure of the proteins, regardless of how we get it. However now we find that we can as a bonus also sample folding pathways at the same time to track HOW the proteins fold and this can later be compared with experiment. If they match it suggests our folding simulations may approximate reality. In order to handle much larger protein sequences than F@H, we obviously have made concessions. We do not use the 'tried and true' MD approach (which has been around for decades) but rather use an extension of our TraDES algorithm to perform a coarse-grained dynamics simulation which takes much larger steps in conformational space. It remains to be seen with the beta testing if this approach can get us to the native fold of the protein, given an ideal scoring function to guide it, but so far we have gotten close (4.5A RMSD is the record so far I believe, from the previous beta - ideally we should get 1-2A which is why we're still trying different algorithms).

If you got lost somewhere above there, you may have to do some reading on the web site to make sense of things :p