Here we go:





:)

Nice! But I beat you!!!!! I got three! Lol

https://boinc.bakerlab.org/rosetta/rah_results.php?UserID=201050

Don't mind me... =]

---

I'm showing 12 lowest RMSD at the moment, but most are a highly suspicious value of "1".
I bet there are thousands more too. just a quirk of the model, no doubt.
I'm not getting too excited about it.

My results

---

I also scored well with nine 1's. The energy levels for the most part were within 25 of the lowest but the best one had an energy level of 1 while the winner had -353.003!

---

Chilean's not the only one who's rocking....and I have a weird double greyed area...what's that mean?

---

The gray is where your value matches the best value. So gray is a good thing, but I agree, there seem to be too many of them and 1.0000 is a suspicious value. I sent an EMail to the Project Team asking about this a few days ago.

---

Rosetta Moderator: Mod.Sense

The 1.000 are indeed wrong. They stem from score-lines that

look like this:

SCORE: 1.#QO 0.000 0.000 0.000 1.#QO 1.#QO 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 632.000 1.#QO 1
.#QO -1.#IO -23.710 0.000 0.000 0.000 0.000 1.#QO 0.000 0.009 0.006 1012630.125
-0.002 -1.000 1.#QO 1.000 1.#QO 1.000 1.#QO 1.000 1.#QO 1.000 160.000 2663.000 S_00001_0000229_00091 99407

the 1.#QO is misinterpreted.
I also don't know where these structures come from. I never got such output on our local machines.

-Oliver

I have 11 greyed now. I do know that I got awesome granted credit from these "real core" work units too, high 200's to mid 300's.

---

The gray is where your value matches the best value. So gray is a good thing, but I agree, there seem to be too many of them and 1.0000 is a suspicious value.

I'm now at an embarrassing 30 low RMSD scores, but is that what we're aiming at or lowest energy? Or a combination of both at the same time?

I've got one lowest energy hiding in there (by quite a long way judging by the graph, though quite a high RMSD). Is that especially good or not?

---

I'm at 12 now, and the one is still greyed in 4 areas. I'm suprised that I'm getting some of the best credit I've had with the new processor, but Rosetta's overall TFlops is not very high right now, not like it has been in the past this summer.

---

I have about 45 now....

---

Dude...you rock. But then you have more than one computer. I only have one. And I have 12 greyed areas now. I think Chilean is going to win the Golden Chromosome Award. 8-)

---

Dude...you rock. But then you have more than one computer. I only have one. And I have 12 greyed areas now. I think Chilean is going to win the Golden Chromosome Award. 8-)

My RAC is pretty low compared to those who administer networks lol.

Oh yeah, I just built a PC with the same CPU (an AMD Athlon x2 7750 OC'ed to 3.1GHz) for my grandpa. Put BOINC in it, since the hardest thing he does is have 3 excels open and a javascript looking at the stock markets =]

---

The gray is where your value matches the best value. So gray is a good thing, but I agree, there seem to be too many of them and 1.0000 is a suspicious value.

I'm now at an embarrassing 30 low RMSD scores, but is that what we're aiming at or lowest energy? Or a combination of both at the same time?

I've got one lowest energy hiding in there (by quite a long way judging by the graph, though quite a high RMSD). Is that especially good or not?

RMSD is a measure of how close your result was to the known structure of the protein, whereas the Low Energy is
a measure of the latent energy of a particular structure.

In theory, the structure with the lowest energy will be the one closest to the native structure (i.e. the lowest
RMSD) as the protein in nature will fold itself into as low an energy structure as it can.

When we are crunching a protein with an unknown structure we cannot use the RMSD, so the only guide is the
Low Energy. So the ultimate goal of the project is to find completely unknown protein structures with the Low
Energy results as the prime key for narrowing down the possible structures from the millions to just a handful.

---

I'm now at an embarrassing 30 low RMSD scores, but is that what we're aiming at or lowest energy? Or a combination of both at the same time?

I've got one lowest energy hiding in there (by quite a long way judging by the graph, though quite a high RMSD). Is that especially good or not?

RMSD is a measure of how close your result was to the known structure of the protein, whereas the Low Energy is a measure of the latent energy of a particular structure.

In theory, the structure with the lowest energy will be the one closest to the native structure (i.e. the lowest RMSD) as the protein in nature will fold itself into as low an energy structure as it can.

When we are crunching a protein with an unknown structure we cannot use the RMSD, so the only guide is the Low Energy. So the ultimate goal of the project is to find completely unknown protein structures with the Low Energy results as the prime key for narrowing down the possible structures from the millions to just a handful.

Thanks for that neat explanation, Murasaki.

So, having a lowest score in the 1st2nd columns, which we're all getting in abundance, isn't what we're aiming at (much as I thought). It's the low energy in the 3rd4th columns we should be getting a little more excited about (I've now got a second one).

Though I suppose a low energy with a high RMSD might indicate a low point that's something very different from the known structure. Might that be even more interesting?

---

So, having a lowest score in the 1st2nd columns, which we're all getting in abundance, isn't what we're aiming at (much as I thought). It's the low energy in the 3rd4th columns we should be getting a little more excited about (I've now got a second one).

Though I suppose a low energy with a high RMSD might indicate a low point that's something very different from the known structure. Might that be even more interesting?

Well, the aim is to get as low an energy and as low an RMSD as possible, so both are important. The RMSD is really a test to see
how well the scientists have designed the process. If we are able to provide them with answers that have both low energy and
low RMSD on a regular basis for known structures, then they know that they should be able to achieve similar results on blind
tests where the RMSD is unknown.

Results with low energy but high RMSD are interesting in that they can suggest several things:

1) We haven't crunched the work units for long enough and there is a low energy structure out there that we haven't hit on yet.
This would be a problem in a blind test as you wouldn't know how long to keep crunching before deciding the result with lowest
energy was the best answer.

2) There is some factor like an inter-molecular bonding force that the model has not considered, so the low energy results are
false. If the project team are able to add in the unknown factor it should improve the quality of future results.

3) The protein has several possible structures with similar energy levels. Further modelling won't help too much, so the project
team has to determine which answer looks best through other means. The project team could compare the results to similar
proteins with known structures and if they find one that is a close match then that suggests the result may be the right one.
Experimentation is another way to determine the answer, but the goal here is to find the right answer without needing to carry
out expensive and time-consuming experiments.

4) The "known structure" found by experimentation may not be exactly correct. The protein has to be in crystal form for the
experiments to be carried out. When we model the protein's natural structure it is possible that we may hit a result more
accurate than what it is possible to find in an experiment.

---

Well, the aim is to get as low an energy and as low an RMSD as possible, so both are important. The RMSD is really a test to see
how well the scientists have designed the process...

I keep forgetting that. Thanks again for the discussion.

---