Reply With QuoteWooHoo! Time to kick some CASP butt!
The CASP5 (http://predictioncenter.llnl.gov) structure
prediction experiment is now underway, and will be proceeding throughout
the course of the summer. Thus, starting towards the end of next week
(Thursday approx.) we will prematurely terminate the present protein we
are working on to begin phase III, on the first CASP target. We will
resume phase Ib, with the next large protein of known structure, once
the CASP 'prediction season' has passed. From the users' point of view,
this simply means an auto-update will occur and RMSDs will no longer
be computed (since the structure of the new protein is not known).
Auto-update should now work for all OSes and versions to the best of our knowledge.
Howard Feldman
WooHoo! Time to kick some CASP butt!
is there a guess/feel for how big these stuctures are in terms of how long they will take?
that is, if a pc is currently generating 50,000 structures a day with the current protein, will this new thing be faster/slower?
Use the right tool for the right job!
Team MacNN is ready to go.
Crunch...Crunch...Crunch
Yeah, that is what I have been waiting for. Once that starts I will be moving all my machines from the ECC contest I am currently working on to help you out with CASP5.
I guess it was good timing to start working on the dfGUI program as this will probably attract more people to the project.
Howard, thanks for posting the results of phase1a, its good to see the results of what people put their effort into.
Jeff.
Amen to that!Originally posted by Digital Parasite
Howard, thanks for posting the results of phase1a, its good to see the results of what people put their effort into.
So will there be anything like the smallest RMSD or such that we will be told? I assume there's some kind of weighing system you guys will be using or will you turn in all the structures you generate? Or would that tip your hand to your competition? Then again, it could strike fear into them!!!! Bwahahahaha!
Oh, and now would be a great time to hire a full-time marketer!
The RMSD is not possible with unknown proteins because there is no known protein to compare it to. There is no difference in structure to derive from because the true structure isn't yet known. Or something like that.
They DO have other 'scoring' methods and that is what they have been looking into and testing with the first phase of the project. I get the impression that they are using the RMSD to get proteins that *should* be pretty close and then are applying thier OTHER scoring methods (which don't rely on prior knowledge of the protein structure) to the protein and seeing how close the other methods are. I didn't ever see this stated as such in the science page/papers, but that is the impression I got.
I just hope the auto-update goes smoothly.
I would assume that the new protein will be the declared CASP5 target (HI0817 from Haemophilus influenzae).Originally posted by FoBoT
is there a guess/feel for how big these stuctures are in terms of how long they will take?
that is, if a pc is currently generating 50,000 structures a day with the current protein, will this new thing be faster/slower?
If so, it will be 182 AA and so will be a lot slower than the current protein.
Of course I could have mis-understood the science here...
Ni!
It isn't only the residue # that affects (effects?) the length of the folding process. The geometry of the protein has a lot to do with it. When the protein 'sample' tries to cross over itself, it has to jump back and try a different route. There are better explanations than this, but the TYPE of protein can make a large impact (based on phase 1a examples) to the speed of processing.
It was my understanding that HI0817 from Haemophilus influenzae, is just one of many different possible CASP 5 targets, and just the first one to be made available. If I understand what I read before correctly, they will choose several of the CASP5 targets and will attempt to predict them, but probably not the whole bunch. I think we would need a whole lot more crunch power to do them all. However if they do well with the ones they choose, then it will bode well for the future and help to validate their methods.
Hopefully we can get some clarification on that.
Yep, all the above is correct. Also some structures we will be predicting with methods other than DF (ones that don't require such resources).
And more crunching power is coming soon we hope.
Howard Feldman
Are the different groups of scientists as competitive as the DF teams ?
Or is it all smoking pipes and cups of tea ?
Regards
Andy
Even more so, if that's possible
Howard Feldman
Ever talked to a research professor or attended a science conference?
Those guys take competition to another level, in an odd passive-agressive kind of way.
I hope Hogue Bioinformatics and the SLRI do very well, but that doesn't mean I don't hope someone just blows the doors off of the entire competition either. It is important work, hopefully the competition has resulted in multiple groups making significant strides and breakthroughs.
One can only hope.
Yeah, but I'd just as soon it was Howard's team that "blows the doors off" - wouldn't mind be a moderately passive part of such.
ROI on a $50,000 personal farm would suddenly look really good. Especially since I'm strongly contemplating doubling it - primarily for this little endeavor...
I could double my whole farm for less than a thousand. Oh wait, I don't have a farm.....Originally posted by Jodie
Yeah, but I'd just as soon it was Howard's team that "blows the doors off" - wouldn't mind be a moderately passive part of such.
ROI on a $50,000 personal farm would suddenly look really good. Especially since I'm strongly contemplating doubling it - primarily for this little endeavor...
Farm? Hmmm. Mine doesn't even qualify as a window box garden. More along the lines of a flower pot.
But a hard working flower pot!
LOL @ Shaktai & Scott
It's all a matter of scale.
Well in that case then I have 4 flower pots.Originally posted by Jodie
LOL @ Shaktai & Scott
It's all a matter of scale.
It's not the size or number of computers you have crunching/folding or whatever you want to call it, wich counts but the fact that your willing to put your resources into any DC project.
Some people will allways have more then others but my 2 comps cost me as much as someones else his/hers 50 comps, scale wise I mean
Yup! My point exactly!
Hey Jodie, I've got a present for you:
OCN Stats at Zeroth Element
I posted a note over at the OCN forum, but I wanted to make sure you knew since you're kicking so much ass.
My Dyyr Dyyryath -
YOU ROCK! (As always)
THANK YOU!!!
You just single-handedly improved my enjoyment of DF by ten-thousand-fold...
From the day I joined TGC I've said time and again that your stats are simply the be-all/end-all of talent.
Well, it warms my heart to see you so happy!
Does this also mean that the uploaded results of the current protein will not be counted after thursday?
When will this be exactly? I have sent approximately 1.5 million strutures home by e-mail for flushing (approx. 75Mb). It would have been a complete waste of time and effort if I wouldn't get credited if I couldn't upload these results before thursday.
In other words, is there a deadline known yet?
To err is human, to mooh bovine!
The news page has an updated deadline, I think I saw Thursday at 12 EST. EST is currently GMT - 5 (I THINK) but don't quote me on any of this.
everyone should be flushing caches NOW in my opinion
don't wait, you will be disappointed if you don't get them in before the change
JUST SEND THEM IN NOW!!!!
Use the right tool for the right job!
Posting Permissions