DF Daily (10/22)

[m0ti]

Hello there, Folders,


Let's take a look at what's going on in the DF world today...

Hmm... not too much... so let's get to the milestone makers:

JetBlack69 makes his 200K and 300K folds!

Congrats, JetBlack69!

Top Folder: TA TheMagicman with 87.4M (though he's stalled at no production. Come on, get some numbers up! PLEASE!).
Best RMS: Kris captures the best fold award, with a 10.19, which is about the 81st best fold!
Best Production: Still Insidious with a 446K, putting him into the 67th most productive folder for the day! Yay!

Notable Producers For the Day:

Insidious
bfour
Kris
m0ti

TeAm Stuff:

Ok, we're now a mere 117M back of 17th. We could make that, like, in a few days if we could get some really big guns or a lot of members. Anyone thinking of buying a rack or farm, now'd be the time to do it (I'll award a permanent to anyone who does...).

Daily Production: Somewhat down at 2.26M, which puts us back a spot into 21st (though just out of 20th by some 8K).

See everyone tomorrow.

 

[Baldy18]

Thanks m0ti.

Wow, JetBlack69 is really flying up the ranks! WTG!

 

[bfour]

way to go JetBlack! Well baldy gettin' closer! As far as making up, if we get the 20 or so other memberson the DF team active.....
we could probably make it up guicker!

 

[JetBlack69]

Thanks everybody. why are some people not active???:frown:

 

[Insidious]

Thanks m0ti

why are some people not active???

Because they have short attention spans.... they hop from project to project in search of fame and glory and leave a trail of disappointed teammates in their wake!

OK, that was harsh, but it is a shame none the less

edit: /me sighs.... OK, the real truth is that it takes a little experimentation to find a project that is just right for a given individual.... Also, some teams need a little help from time to time *cough* *cough* and the guns will join with them to get them going.

 

[bfour]

I guess I don't have a short attention span! Cracked RC5-64 over 3 years, ended up #24 or 23 on the team of over 3000! I just feel more good can come from crunching proteins than most other projects, that may help out with something in my lifetime!

 

[Robor]

I hear ya'll talking and I know I'm one of the slackers but I'm in a MTR race right now so I can't afford to put my clients back to DF just yet. Don't worry though. We've got a nice lead over BBR now and the team has upped production since the end of RC5 so I will be switching some power back to DF - just not right now.

Thanks m0ti!

 

[Insidious]

 

[CrazySaint]

I'm currently trying out and looking into various DC projects right now in search of a DC "home" - presently trying to decide between crunching for DF or "trading CPUs" with a guy on the SETI team (I've got a P4, he's got an Athlon) and crunching for ECCp. My CPU is running at 2.7GHz (OC'd 1.8A), with it set to use up to 150MB, based on a very rough estimate (I timed a very small part of one set and extrapolated over a period of 24 hours), I figure I could output something like 55-65k per day if I left it running 24/7. I'm actually approx. half way through my first DF set right now. If I take him up on his offer to "trade CPUs", I'll probably come back to this after the SETI races on Nov. 5.

 

[Mloot]

Crazysaint: Your rig should do much better than 60K/day. That's what I get on my home rig w/1.7G celeron(stock speed) and 512mb pc2100 (using 150mb of RAM). Your P4 should smoke on DF.

 

[CrazySaint]

Originally posted by: Mloot
Crazysaint: Your rig should do much better than 60K/day. That's what I get on my home rig w/1.7G celeron(stock speed) and 512mb pc2100 (using 150mb of RAM). Your P4 should smoke on DF.

Looks like my estimate was rougher than I thought. Its been about an hour since my last post where I said I was about half way through my first set, well now I'm about half way through my second set, so based on that sample (1 5k set/hour), my production should be closer to 120k/day

 

[CrazySaint]

Wow, some of these proteins are big! I had one that took several minutes to calculate.

*Currently calculating structure #10124*

 

[FoBoT]

Originally posted by: CrazySaint
I had one that took several minutes to calculate.

the ones that take a loooong time are the "good" ones, in terms of the science
they cut your production of garbage (the fast ones) but the ones that have the low RMS values (which is the whole point, to get a really low RMS value protein) take a looong time, like 20-40 minutes roughly

so long caluculations are a good thing

 

[CrazySaint]

Originally posted by: FoBoT

Originally posted by: CrazySaint
I had one that took several minutes to calculate.

the ones that take a loooong time are the "good" ones, in terms of the science
they cut your production of garbage (the fast ones) but the ones that have the low RMS values (which is the whole point, to get a really low RMS value protein) take a looong time, like 20-40 minutes roughly

so long caluculations are a good thing

So what exactly is RMS and what determines how high/low it is? Distributedfolding.com is definitely a page in need of a FAQ.

 

[m0ti]

Actually, quite a bit of information is available on the DF site.

Take a look at the "About" link. As well, there's plenty of great information floating around in the forums (at Free DC), since Howard Feldman, one of the project's principle organizers (and programmers), posts very, very often.

RMS is Root Mean Square. It's a method by which they measure the difference between the protein generated and what it should look like (the RMS of the distances from where things should be to where they are). While we were participating in getting our CASP 5 submissions ready, this was called pseudo-energy, since we didn't know the actual structure. They measured a type of energy according to some scoring function, and ones with less energy tend to be more likely to show up. For the actual submissions, many of the best structures were further tested (think harder to calculate scoring functions used. Or many of them) to determine which of those were the best of the best.

 

[FoBoT]

ouch, you made my brain hurt, i don't know much about the science part, i am just a simple minded moderate stats whore

but here is some cut/paste stuff that pertains

Generally, a bigger protein will have larger RMSD values when compared to random conformers of the same length and sequence than a small protein would. However, an RMSD of 6A or less for a protein of fifty or more residues indicates that the general topology is correct. At this level of resolution, crude scoring functions can be used successfully to help identify the native fold.

We have estimated that for a typical medium-sized protein of about 120 residues, by sampling on the order of 1 - 10 billion probabilistic conformers a structure within this 6A tolerance could be sampled, and would probably have the correct topology (i.e. fold) and one of the lowest energies out of the structures generated. If this hypothesis is correct, this means that the structure of almost any 120 residue or shorter protein could be predicted ab initio, i.e. without any prior knowledge about the protein.

By harnessing the combined computational power of computers across the internet running our Foldtraj algorithm, which can generate a single conformer on a desktop PC in about 2-3 seconds, we hope to be able to generate up to 10 billion conformers of a series of proteins in a matter of weeks and test this hypothesis. On a single machine, this computation would take just under 10 million hours (1000 years). However, if run on just 10,000 CPUs simultaneously, this would cut the time down to a month!

This would allow the structures of newly discovered proteins with unknown functions to be predicted. Once the structure is known, the function can often be inferred by its shape or similarity to other known protein structures. With the recent completion of the sequencing of the human genome, we now have a pool of several thousand proteins, found in the human body, whose purposes are completely unknown to us. Many of these could be related to diseases which are subject to intense study and whose causes are presently unknown, such as Alzheimer's or certain type of cancer. Learning the structures and functions of all the proteins encoded in the human genome is the first step towards understanding these diseases better, and thus finding a cure.

link to the page with that stuff above

ok, my brain is mush now :Q

in simpler terms, the lower the RMS value, the "better" the structure

the stats page on the DF home page lists the ten best structures found so far for the current protein
while the big producers rack up millions upon millions of "garbage" stuctures, it is very common for a person crunching with a single pc or just a couple to be on the top ten structures list, everybody can contribute in this project

the current top ten structures are

Best structure so far AnalAlbatross Number 1 is our rightful place Ars Technica Team Stir Fry 8.88
Wed Oct 16 23:30:30 2002

Runners up: - - - - -
#2. happyboy99 Singapore Singapore 9.03
Sat Oct 19 23:00:25 2002 -
#3. DGROMScom Here comes the XP2400 Train Overclockers Australia 9.20
Fri Oct 18 02:40:25 2002 -
#4. Andy Read none Team Read 9.21
Tue Oct 22 12:40:34 2002 -
#5. Raccoon ProCooling ProCooling 9.22
Wed Oct 16 18:10:26 2002 -
#6. devzero the path of least resistance FreeDC 9.37
Fri Oct 18 07:40:26 2002 -
#7. mr sean Fear & Loathing Inc. Team Endeavor 9.39
Sat Oct 19 01:00:26 2002 -
#8. David ProteinDesign dot com Proteindesign.com 9.40
Sun Oct 20 19:10:34 2002 -
#9. Welnic Asylum Research Ars Technica Team Stir Fry 9.43
Tue Oct 22 18:50:28 2002 -
#10. Conydd France Alliance Francophone 9.43
Thu Oct 17 10:40:32 2002

well, it looks better in the html table, anyway , as you can see for the current protein, AnalAlbatross of Ars Technica Team Stir Fry has the lowest RMS structure with a 8.88
howard, the guy behind the project has said that for this protein, he hopes we can produce a structure with an RMS value close to 6A within the 10 billion structure sample

 

[CrazySaint]

Thanks for the cut/pastes guys, I somehow managed to completely "not see" the menu that says "About Downloads High-Flyers News" and only saw the one that said "Teams Known Bugs and Fixes Discussion Forum (now hosted at Free-DC)" D'oh!

*Will finish structure # 15k in a few minutes*

 

[DJSnairdA]

Thanks m0ti!

 

[bfour]

checking stats this morning looks like we gained a few more members!