Help a guy out.

[relic]

Hello All.

It's that time of year again, and the [H] DC team is evaluating projects for participation.

I'm swamped doing testing, so I was hoping that some of my good friends could get me jump started on the DF project. I haven't run it in a while, I'm sure things have changed.

I'm going to start running DF again tonight. What I could use is some info to help me fast track this eval. What is great about the project? What needs fixing?

The web page is pretty sparse...is there a FAQ?

I'm looking for things like project goals, milestones and accomplishments, sponsors, etc. Where should I look?

Much Thanks,
-Steve

[KWSN_Millennium2001Guy]

http://www.distributedfolding.org/about.html

[FoBoT]

in its current form, DF is extremely sneakernet friendly

however, howard has posted some info that in order to make the science better, it may be necessary (fairly soon) to require the client to touch the server somewhat frequently

this will lower the sneakernet friendliness of DF by several magnitudes :/

[GenomeX]

Originally posted by relic
Hello All.

<snip>

I'm looking for things like project goals, milestones and accomplishments, sponsors, etc. Where should I look?

Much Thanks,
-Steve

A goal? To beat OCAU

This has nowhere near the bandwidth restriction F@H has. Much better for dialupers which is why I stopped F@H.

Its good to see [H] taking some interest.

[MAD-ness]

http://www.teamstirfry.net/faq.php

It is out of date and not as complete as I would like, but it is a pretty good place to get some questions answered. Worth a look.

BTW, I can't help but mention that if you guys had treated the [H]ard crunchers who were trying to run the Distributed Folding client better early on that you would have a lot less work to do now.

Sorry, couldn't help myself.

Good luck though, we always look forward to additional firepower joining the project and additional teams to compete with.

[relic]

Originally posted by MAD-ness
BTW, I can't help but mention that if you guys had treated the [H]ard crunchers who were trying to run the Distributed Folding client better early on that you would have a lot less work to do now.

Currently we are evaluating DF to see if it we should include it in the team vote.

Rude people like yourself associated with the DF project go a long way toward helping me form an opinion.

[whilden@ars]

-Relic-

Mad-Ness isnt a rude person, I think he was sticking up for me a bit (I think we both remember the thread where I tried to Defend DF in your forum)! Anyways name calling isnt going to solve anything and the good folks at DF would love to have some of your teams cycles! This DC project is Top Tier and it should be a no brainer [H] project!

Anyways, as always I give this client my 3 thumbs up!

[relic]

Originally posted by whilden@ars
-Relic-

Mad-Ness isnt a rude person, I think he was sticking up for me a bit (I think we both remember the thread where I tried to Defend DF in your forum)! Anyways name calling isnt going to solve anything and the good folks at DF would love to have some of your teams cycles! This DC project is Top Tier and it should be a no brainer [H] project!

Anyways, as always I give this client my 3 thumbs up!

Rude is as rude does.
I have no use for a punk like Mad-Ness who attacks me and my team while I'm asking for help.

Regardless, [H] is evaluating new projects. DF is being considered. (Entered for consideration by myself, I might add). Now is the time to post your DF comments for [H] consumption, if you so desire.

I would welcome any comments that aid in that evaluation.

Regards,
relic

[whilden@ars]

Relic

Mad-Nes is very similiar with Pointwood and these two guys operate along the same line, I dought you would ever call pointwood a punk!

Once again I ask nicely that you dont call him or other names, its not fair to the project that your trying to evaluate!

He gave you a link to a decent FAQ, run with that and let it be! Anyways, have a good day!

In respone to pointwood's post below: Let's just move on with this issue, mad-ness I understand what you were getting at, its a sore subject with some so lets just move on! I think we can all agree on that! Mad-ness would want nothing better for the [H] to put their cycles here, I can guarantee that!

[pointwood]

Originally posted by MAD-ness
http://www.teamstirfry.net/faq.php

It is out of date and not as complete as I would like, but it is a pretty good place to get some questions answered. Worth a look.

Send me an updated FAQ and I'll upload it right away

BTW, I can't help but mention that if you guys had treated the [H]ard crunchers who were trying to run the Distributed Folding client better early on that you would have a lot less work to do now.

Sorry, couldn't help myself.

I'm sorry too, I see no reason to post it

Shouldn't we just forget about it and welcome HardOCP if they decides to join in a big way? (well, relatively big - they have no chance against TSF )
(Think that will make them join? )

UPDATE: /me agrees with whildens update

[relic]

Originally posted by whilden@ars
........Once again I ask nicely that you dont call him or other names, its not fair to the project that your trying to evaluate!.......

Please read the first post of the thread and then check to see who took the first swing. I reserve the right to retaliate when attacked, and I stand by my comments.

Of course, we're also evaluating the "community" and the attitude of the participants vs other projects we're considering. It's all part of the process.

So. Is there some kind-hearted DFer who would like to provide helpful information to aid in our evaluation of the project? Is the FAQ too far out of date, should I look at other sources of information? If so, where?

Thanks,
relic

[Aegion]

Originally posted by relic

So. Is there some kind-hearted DFer who would like to provide helpful information to aid in our evaluation of the project? Is the FAQ too far out of date, should I look at other sources of information? If so, where?
Thanks,
relic

Alot of detailed information on the science can be found here. The link leads to an official site involved with the project with lots of detailed information on the science and what the project software does.

[runestar]

relic,

Without taking any sides here, it takes a man to fight back... but the true mark of a gentleman is to know when to step away. There's enough retailiation in this world as it is.

That said, here's my two-pence worth on running this project. First, it runs well on a wide variety of machines, speeds, and OSes. We even have some people run this on Pentiums. For a maximum size of 10,000 structures, it takes maybe a few hours on a sub-1Ghz machine. Of course the proteins vary in size so sometimes you get "fast" folding and sometimes you get "slower" folding.

Second, being a relatively young project, its also fairly easy to move up the charts unlike SETI where you could have ten fast machines and barely make a dent in the charts.

Third, some really nice clients, featuring both a virtual ASCII representation of the protein in the command-line client and a brand-new screensaver which allows you to interact with the protein. Interaction includes 3D (x,y,z axes) rotation, zoom in/out, moving around the screen, actual percentage of the screen to take up, and more. For the ASCII client, there is the infamous dfGUI for Windows which allows you to control the various aspects that normally you would have to manually edit the command-line switches.

Fourth, and perhaps one of the best reasons is the interaction with founder Howard Feldman whose presence is almost daily here in the forums. I've seen him post here even on the weekends. He interacts with the community and always takes suggestions and requests... although that doesn't mean he is going to implement every single one.

Best,

RuneStar&#189;

[runestar]

P.S. I use to participate in SETI but I grew tired of it and started experimenting with other DC projects. I periodically would run SETI@home, but I quit that permanently when I learned about their refusal to address the rampant cheating.

It took an article in ZDNet and a petition by the community for them to find some time to address it. If Howard ever said something to the effect that he would address the cheating (there isn't any by the way just by shear virtue it's almost impossible) when he had the spare time, this project would be dead. Those were the words David Anderson, founder of the SETI@home project used.

Well, apparently, a lot of SETI crunchers have decided that they too would take that same philosphy... we'll crunch SETI when we find the time.


And that's another point for this project, that DF is generally immune to cheating and hacking. Nothing is totally secure, but people have been quietly testing the project for security holes long before even I appeared.

Hope all this is useful for all of you. BTW, as far as I know, no other project has the same level of interaction with the project founders as DF with Howard Feldman.


Best,

RuneStar&#189;

[pointwood]

You should not forget to read the readme that comes with the client. It contains *a lot* of important and useful information.

Otherwise I think most relevant info has been linked to.

If you got more questions, then please don't hesitate to ask and it will be answered.

Ps. Please don't judge MAD-ness (or this community - browse this forum if you like and see if you can find a lot of flamewars, if you're very lucky, you'll find one since very few exists) based on that post alone. He has probably not posted again yet because of problems with his internet connection (he *always* have problems with his connection ), but my point is that MAD-ness isn't normally the person that attacks others like that. In the fairly long time I've known him from TSF, this is the first time I've seen him post anything like that. Everybody makes mistakes.

[relic]

Originally posted by Aegion
Alot of detailed information on the science can be found here. The link leads to an official site involved with the project with lots of detailed information on the science and what the project software does.

Thanks, good link.

Something I don't understand is "why?" Eventually computer modelling will replace tedious laboratory procedures? I guess my question is "What's the purpose of DF?"

[relic]

Originally posted by runestar
Hope all this is useful for all of you. BTW, as far as I know, no other project has the same level of interaction with the project founders as DF with Howard Feldman.

Actually very useful, thank you.

PS. Howard is more than welcome to stop by the [H]ardforum and provide the same level of interaction as the other project administrators.

You may consider this a personal invitation to Howard to pimp DF on the [H]ard DC forum.

[relic]

Originally posted by runestar
relic,

Without taking any sides here, it takes a man to fight back... but the true mark of a gentleman is to know when to step away. There's enough retailiation in this world as it is.

I follow the tenfold rule of return. Good or bad. It's effective...and I'm all about "effective". I have exactly zero tolerance for a**holes.

That said, here's my two-pence worth on running this project. First, it runs well on a wide variety of machines, speeds, and OSes. We even have some people run this on Pentiums. For a maximum size of 10,000 structures, it takes maybe a few hours on a sub-1Ghz machine. Of course the proteins vary in size so sometimes you get "fast" folding and sometimes you get "slower" folding.

Cool, thanks.

Second, being a relatively young project, its also fairly easy to move up the charts unlike SETI where you could have ten fast machines and barely make a dent in the charts.

Well, we're more of a "come from far behind team" but it would be nice to help out a new worthy project. We're evaluating the "worthy" part.

Third, some really nice clients, featuring both a virtual ASCII representation of the protein in the command-line client and a brand-new screensaver which allows you to interact with the protein. Interaction includes 3D (x,y,z axes) rotation, zoom in/out, moving around the screen, actual percentage of the screen to take up, and more. For the ASCII client, there is the infamous dfGUI for Windows which allows you to control the various aspects that normally you would have to manually edit the command-line switches.

Eye candy and third party support. Good stuff.

Thanks and Regards,
relic

[Aegion]

Originally posted by relic
Thanks, good link.

Something I don't understand is "why?" Eventually computer modelling will replace tedious laboratory procedures? I guess my question is "What's the purpose of DF?"

Ok, the current method of discovering the shape of a folded protein involves X-ray Crystallography or Nuclear Magnetic Resonance Spectroscopy, often taking six months to a year to discover the shape of a single protein once it has completed the folding process. Certain types of proteins are not amenable to being solved by either of these methods. Distributed Folding offers the potential to accurately predict the shape of the same protein in about a week or so much more cheaply, and potentially allows us to discover the shape of proteins that we would not otherwise be able to determine.

This gives researchers attempting to create medicines much more abundant data on different proteins that may be relevant to their research and makes the cost of discovering the shape of a particular protein far less. This makes it commercially viable to research drugs for particular diseases where this might currently not be the case and potentially lowers the cost of new drugs since the research cost is lower.

[IronBits]

I can't believe no one mentioned the number one reason to run any project.
/me snickers
STATS
http://stats.zerothelement.com/cgi-b...ender-teams.pl
Best on the planet

[relic]

Thanks, Aegion, has there been any timelines suggested as to when we get out of the "validating the methodology" phase?

Or perhaps a better way of saying that is "When do we stop duplicating lab data of known proteins and start doing new unsolved structures?"

Is there a plan to end the validation and move the project into a production environment?

[relic]

Originally posted by IronBits
I can't believe no one mentioned the number one reason to run any project.
/me snickers
STATS
http://stats.zerothelement.com/cgi-b...ender-teams.pl
Best on the planet

Amen, Dyy<tab> is a genius.

[Aegion]

Originally posted by relic
Thanks, Aegion, has there been any timelines suggested as to when we get out of the "validating the methodology" phase?

Or perhaps a better way of saying that is "When do we stop duplicating lab data of known proteins and start doing new unsolved structures?"

Is there a plan to end the validation and move the project into a production environment?

We already did so when we participated in the CASP 5 trials some time ago. Presumably when we switch to the new algorithm and test it for some time, we will start working on unknown proteins again. The knowledge of how to accurately predict protein folding is a huge advancement on its own. While this schedule has been slightly altered since it has been set, the general schedule for protein structure predicition can be found here.

Obviously Howard could give you additional info if he decides to chime in here.

[Brian the Fist]

Originally posted by relic

You may consider this a personal invitation to Howard to pimp DF on the [H]ard DC forum.

while I would love to be your pimp :shocked:, this forum keeps me busy enough. This is the official forum of the project and I read and address all concerns posted here (and only here). That said, I will now address your concerns

As mentioned, the CASP5 experiment served as a 'blind test' of our method by attempting to predict proteins of unknown structure, and then comparing them after the structure was solved (see http://predictioncenter.llnl.gov/casp5 for more info about CASP). Our main goal though is algorithm and methods development. We wish to develop and improve our structure sampling and scoring methods. DF allows us to rapidly test new ideas and scoring functions on huge samples of structures, and determine what works and what doesn't basically. In a sense we are searching for the 'holy grail' of structure prediction. That combination of sampling approach (i.e. how we generate putative structures) and scoring function (i.e. how we pick the most correct structure from the set of putative ones) that lead to the true fold for most/all proteins. Thus as we progress we will be testing different methods. These must generally be done on known proteins, in order to determine what works and what doesn't. Ultimately as our methods improve, they can eventually be used (after publication) by others to predict novel protein folds for their own purpose. Presently, no one can predict 3-D protein structures 'ab initio' with any degree of confidence however. But this is our ultimate goal. Our algorithm is still in its infancy and tehre is still lots of room for improvement, and in fact we are in the process right now of designing a brand new, smarter sampling approach (being discussed in another thread).

There's also a lot of good info inthe Educational section of this forum by the way. If you have any other questions I'd be happy to answer them, or direct you to teh answer if its already been asked.

[runestar]

Just my own note here...


Validation, as it was put, is an important part of the protein folding. Howard et crew aim to make sure that their protein folding algorithim is the best it can be. I'm not as familar with the science as some of the others here, but if they can create structures close to existing known structures, that's a good sign. If not, then they obviously need to tweak things more.

The more firepower we get NOW, the faster we can get out of the "validation" stages. With more computing power, we can run through the sets faster and thus Howard et al. can evaluate those results and make the necessary changes.

Plus, if you start off now, you can brag to all the new people later on you were here in the beginning.

I know one thing for sure, if Howard can get some more crew as this project expands and keep up that visibility between himself and the crew, F@H is going to have a serious rival.

TTFN,

RuneStar&#189;

P.S. One other plus for that project I forgot was that its non-commercial, that is it is not run by a commercial organization. This may or may not be something some care about, but I know many DCers (myself included) have a strong preference for participating non-profit projects.


P.P.S. To relic:
regarding toleration... I understand where you are coming from, but I found sometimes its best not to "retaliate" but rather calmly respond to it. IMHO, that puts you at a much higher level than the other person. Best, RS&#189;

[MAD-ness]

I sometimes forget that being part of a team means that I represent said team.

I had no intention of starting a flame war or of discouraging the [H]ordes from participating in the Distributed Folding project.

Quite the opposite, in fact. I would love to see [H]ardOCP formally add the Distributed Folding project as an official [H]ard DC project. If there is anything that I can do to help you become comfortable with the client or the project, I would be more than willing to do so.

I do not have a personal history or grudge with [H]ardOCP. What I do have an issue with is the treatment that various members of the [H]ardOCP DC community received much earlier in the life of the project when they were attempting to get the [H]ardOCP DF team off of the ground. I read the DF threads at the [H]ard forum and even made several posts, if I remember correctly (which is always a question mark).

I meant no disrespect to relic or [H]ardOCP but neither do I have any respect for people who treat thier teammates and friends the way I witnessed the early DF participants at [H] be treated.

All of the (pointless) politics aside, I feel strongly that the DF project is both worthwhile and well managed as do many of the other more active participants who have posted here offering help and advice.

We anxiously await the competition that the [H]ordes can provide.

[pointwood]

Originally posted by IronBits
I can't believe no one mentioned the number one reason to run any project.
/me snickers
STATS
http://stats.zerothelement.com/cgi-b...ender-teams.pl
Best on the planet

Well, actually I have already told them on their own forum:

Originally posted by pointwood
In regards to stats, you, of course, have the always reliable Statsman stats

Furthermore, Dyyryath (I believe several of you know who he is), have created some, IMHO, really cool stats right here:
http://stats.zerothelement.com/cgi-b...ender-teams.pl