I've got my first graphic workunit now :) i saw it in the queue, so I suspended all other workunits, just to see the graphic one :)

looks amazing, although I haven't got a clue what exactly I'm looking at in the scientific way.

---

I've got my first graphic workunit now :) i saw it in the queue, so I suspended all other workunits, just to see the graphic one :)

looks amazing, although I haven't got a clue what exactly I'm looking at in the scientific way.

Well, this reply will do one of two things. It'll make me look like a candidate for David Baker's team, or it'll make me look like a complete idiot. :)

It's also important to note that all I've seen is the graphic above posted by Jack, so I'm really trying to interpret that.

That being said, my understanding is that the long colored "string" that wiggles round is the protein we care about at the moment. It's wiggling round, and as it does so the two graphs at the side change.

One shows the energy that's contained in the molecule, we'd like to get that as low as possible. The other shows the RMSD, which I'm not sure exactly what that stands for. My GUESS, and I must emphsize that this is only an educated guess, is that it stands for Root Mean Square Distance. Whatever the case, I believe that it's a metric for how spread out (?) the molecule is. Again, smaller is better.

Each time we get a pair of positions in those two graphs, they map to the third graph, and add a spot. The ideal situation is to get a spot as close as possible to the lower left corner, because that's where the molecule will land in reality. If we get there, then we should have found the shape that this molecule assumes, and doing that is a large part of what we're after.

OK, so that is my (short) paper on the subject, I'm curious to know how it'll get graded. :)

---

Lucky me, I got some WU with graphics.
It looks great!

There are two suggestion:
On a second line of text from bottom, there is a name of WU and "Accepted Energy". But as WU name is long string, it goes much too close to "Accepted Energy"; they are touching each other. Hope that it's clear...sorry if is't not smart description.
Second, the very bottom line that show Rosetta@Home, and User: name, team name. Well, there is a lot of space on right side. I known i will be filled when name is longer than mine (or perhaps longer team description than Czech National Team). How about use justification of user to the center and team from the right? If this was applyed to line with stage (from left), Step in middle, and Accepted RMDS in right, same on second line, it may solve the problem of Accepted energy being to close to WU name...

Not that this is critization - all from Rosetta team are doing great and we appreaciate your regular visit and actually posting on the forum. It's important for every project to keep in touch with particiapnts, albeit some project doesn't seems to realize so.

---

I've got my first graphic workunit now :) i saw it in the queue, so I suspended all other workunits, just to see the graphic one :)

looks amazing, although I haven't got a clue what exactly I'm looking at in the scientific way.

Well, this reply will do one of two things. It'll make me look like a candidate for David Baker's team, or it'll make me look like a complete idiot. :)

It's also important to note that all I've seen is the graphic above posted by Jack, so I'm really trying to interpret that.

That being said, my understanding is that the long colored "string" that wiggles round is the protein we care about at the moment. It's wiggling round, and as it does so the two graphs at the side change.

One shows the energy that's contained in the molecule, we'd like to get that as low as possible. The other shows the RMSD, which I'm not sure exactly what that stands for. My GUESS, and I must emphsize that this is only an educated guess, is that it stands for Root Mean Square Distance. Whatever the case, I believe that it's a metric for how spread out (?) the molecule is. Again, smaller is better.

Each time we get a pair of positions in those two graphs, they map to the third graph, and add a spot. The ideal situation is to get a spot as close as possible to the lower left corner, because that's where the molecule will land in reality. If we get there, then we should have found the shape that this molecule assumes, and doing that is a large part of what we're after.

OK, so that is my (short) paper on the subject, I'm curious to know how it'll get graded. :)

A+ by me. I've got one question though, in the RMSD square there's nothing, it's empty. And the box marked Native has nothing in it, just a question mark. Is it because the work unit is just at 20%?

---

Ok, it looks like there still may be a bug to iron out with the Graphics unit beta. I was viewing the graphics for a unit which was about 20% complete. (I had received a graphics unit sometime previously and that one had processed the protein just fine.) Suddenly the unit terminated and another one was downloaded from the server at which point the program proceeded normally. The error message sequences was as follows.

11/22/2005 5:50:32 AM|rosetta@home|Unrecoverable error for result 1dcj__abrelax_rand_len10_jit02_omega_sim_graphics_00353_0 ( - exit code -1073741819 (0xc0000005))
11/22/2005 5:50:32 AM||request_reschedule_cpus: process exited
11/22/2005 5:50:33 AM|rosetta@home|Deferring communication with project for 1 minutes and 0 seconds
11/22/2005 5:50:33 AM|rosetta@home|Computation for result 1dcj__abrelax_rand_len10_jit02_omega_sim_graphics_00353_0 finished
11/22/2005 5:51:32 AM|rosetta@home|Sending scheduler request to https://boinc.bakerlab.org/rosetta_cgi/cgi
11/22/2005 5:51:32 AM|rosetta@home|Requesting 8640 seconds of work, returning 1 results

---

I've got my first graphic workunit now :) i saw it in the queue, so I suspended all other workunits, just to see the graphic one :)

looks amazing, although I haven't got a clue what exactly I'm looking at in the scientific way.

Well, this reply will do one of two things. It'll make me look like a candidate for David Baker's team, or it'll make me look like a complete idiot. :)

It's also important to note that all I've seen is the graphic above posted by Jack, so I'm really trying to interpret that.

That being said, my understanding is that the long colored "string" that wiggles round is the protein we care about at the moment. It's wiggling round, and as it does so the two graphs at the side change.

One shows the energy that's contained in the molecule, we'd like to get that as low as possible. The other shows the RMSD, which I'm not sure exactly what that stands for. My GUESS, and I must emphsize that this is only an educated guess, is that it stands for Root Mean Square Distance. Whatever the case, I believe that it's a metric for how spread out (?) the molecule is. Again, smaller is better.

Each time we get a pair of positions in those two graphs, they map to the third graph, and add a spot. The ideal situation is to get a spot as close as possible to the lower left corner, because that's where the molecule will land in reality. If we get there, then we should have found the shape that this molecule assumes, and doing that is a large part of what we're after.

OK, so that is my (short) paper on the subject, I'm curious to know how it'll get graded. :)

A+ by me. I've got one question though, in the RMSD square there's nothing, it's empty. And the box marked Native has nothing in it, just a question mark. Is it because the work unit is just at 20%?

I agree on the grade!
in some of the jobs we sent out last night the file containing the native (correct) structure was not properly read, so while the folding calculation proceeded normally, the RMSD (how similar the predicted structure is to the native structure) was not computed, and there was nothing to display in the NATIVE box, sorry about this--it was my fault! this problem will be fixed in the next batch of jobs that go out today so you will get to see Jack's production in all of its glory!

---

I got one WU with graphics today and was able to see the last 2 minutes of it coming back from lunch. Awsome! I'm sure with all the inputs from rosetteers that Jack will be able to impress us even more.

---

Lucky me, I got some WU with graphics.
It looks great!

There are two suggestion:

We've already fixed these problems in next iteration of the graphics. Hopefully they'll be sent out soon. I apologize again for the delay!

You'll see in this next round that we're trying include a lot of information down there, and if people have other ideas about how to lay it out, we'll be glad to hear it!

ok, after 2 succesfull graphics WUs couple of days ago i recieved my 3rd late yesterday (and suspended it to not miss it while i am not here :=)), allowed it to start when i got here, all was running fine for some minutes. then suddenly the graphics window disappeared and the boincmanager was totally unresponsive, rest of system was still responding as usual. error messages were as following:

22/11/2005 19:25:34||Couldn't write state file: system rename
22/11/2005 19:26:34|climateprediction.net|Result 3c76_000177345_0 exited with zero status but no 'finished' file
22/11/2005 19:26:34|climateprediction.net|If this happens repeatedly you may need to reset the project.
22/11/2005 19:26:34|Einstein@Home|Result l1_0631.5__0631.7_0.1_T10_S4lC_0 exited with zero status but no 'finished' file
22/11/2005 19:26:34|Einstein@Home|If this happens repeatedly you may need to reset the project.
22/11/2005 19:26:34|rosetta@home|Unrecoverable error for result 1hz6A_abrelaxmode_random_length20_jitter02_omega_graphics_07160_0 ( - exit code -164 (0xffffff5c))
22/11/2005 19:26:34||request_reschedule_cpus: process exited
22/11/2005 19:26:34|rosetta@home|Computation for result 1hz6A_abrelaxmode_random_length20_jitter02_omega_graphics_07160_0 finished

---

still waiting for my first grapic unit... mabey i dont have enough computers running on this project..

---

Though I'm not positive, as the native structure is a bit hard to get a 3D sence of, and I was wondering if you could rotate the native structure slowly to help this?

But first impressions is that:
In 1hz6A_abrelaxmode_random_length20_jitter02_omega_graphics_05110_0 about a third of the way through the WU the structure was almost right but then flipped, this also happened with 1hz6A_abrelaxmode_random_length20_jitter02_omega_graphics_05111_0 but with it, it was the very first structure?
It almost looks like if you did a "full atom relax" at these points you would get closer to the native stucture, instead of going futher on in the ab initio(sp?) stage.

The graphics doen't appear to change the WU time at all, the only graphic unit I've completed (and I watched it all the way through) was about in the middle (timewise) of the same type non-graphic WU's, the second I've only done 6 minuets on so time will tell.

Great job Jack, I hope the David's are paying you well :)

---

Though I'm not positive, as the native structure is a bit hard to get a 3D sence of, and I was wondering if you could rotate the native structure slowly to help this?

Rotating the native is on the list of things to do. Perhaps even rotating rotating everything interactively...

In the meantime, you can actually look at the native yourself in as much detail as you like by downloading the crystal structure from the RCSB Protein DataBank. Here's the page for 1hz6. There are many, many free viewers for PDB files. Including the oldie but goodie rasmol and the fancy but still slightly beta pymol.

But first impressions is that:
In 1hz6A_abrelaxmode_random_length20_jitter02_omega_graphics_05110_0 about a third of the way through the WU the structure was almost right but then flipped, this also happened with 1hz6A_abrelaxmode_random_length20_jitter02_omega_graphics_05111_0 but with it, it was the very first structure?
It almost looks like if you did a "full atom relax" at these points you would get closer to the native stucture, instead of going futher on in the ab initio(sp?) stage.

These observations are great. The thing about the ab initio stage is that it can't always tell when it has found a good structure. It is designed to rapidly, but approximately, search through the large number of possibilities. Then we rely on the the full atom relaxation to see if the best guess from that round of ab initio is actually any good.

The graphics doen't appear to change the WU time at all, the only graphic unit I've completed (and I watched it all the way through) was about in the middle (timewise) of the same type non-graphic WU's, the second I've only done 6 minuets on so time will tell.

Great job Jack, I hope the David's are paying you well :)

Thanks for the positive feedback!

still waiting for my first grapic unit... mabey i dont have enough computers running on this project..

Yeah. Me too. I'm getting jealous of the other crunchers who have had the pleasure of observing the graphics.

Are they only available for certain platforms? My windows box is old and slow and can take up to 24 hours to crunch the longer running workunits. My Linux machine is much faster and I think it takes a bit over 3 hours for the longer running ones. Are they only available for Windows perhaps? I didn't think so from what I've read here but you never know.

Charlie

---

-Charlie

Though I'm not positive, as the native structure is a bit hard to get a 3D sence of, and I was wondering if you could rotate the native structure slowly to help this?

My initial reaction to the graphics? "Ohmigosh, that's sweet!"

However, after looking at it for a bit, I agree with the above comment. The native structure was misaligned with respect to the other three in the one graphic WU I've had to date. I was thinking more in terms of having the WindowProc for the screensaver catch various mouse messages, and by doing so let us drag the mouse to rotate the native structure.

While you're at it, depending on where the mouse is when the left (or right) button goes down, you could also allow scaling of the energy and RMSD graphs, and implicitly the "combined" graph lower right.

I don't know it it would make sense to be able to scale these two along the time axis (horizontal for energy, vertical for RMSD), but there are times when both these could be scaled on the energy and RMSD axes respectively.

These two images show the two problems:





In both cases, you can see how the native isn't aligned correctly, this is most noticeable on the first. I'd say that you'd want to rotate it about 180 degrees about the x axis. Also, both of these show the scaling problem: in the relax phase, the RMSD hardly changes, so it might make sense to zoom that in, while on the second, the enegry has gone off the scale in a couple of places, making a zoom out make sense.

Of course, RMSD only makes sense when we know what the native form is, when we start doing real work, this may not always be available .....

---

in some of the jobs we sent out last night the file containing the native (correct) structure was not properly read, so while the folding calculation proceeded normally, the RMSD (how similar the predicted structure is to the native structure) was not computed, and there was nothing to display in the NATIVE box, sorry about this--it was my fault! this problem will be fixed in the next batch of jobs that go out today so you will get to see Jack's production in all of its glory!

This comment raises an interesting question.

Right now, we're refining the algorithm, and working with proteins whose native structure is already known. This is obviously a good choice at this stage, since it helps determine how well the algorithm works.

However when we go live, and start doing research (e.g. working on a cure for HIV), will we know ahead of time the structure we're after? I tend to think we won't, so does this mean that the Native and RMSD portions of the display will be blank most of the time we're in production?

---

However when we go live, and start doing research (e.g. working on a cure for HIV), will we know ahead of time the structure we're after? I tend to think we won't, so does this mean that the Native and RMSD portions of the display will be blank most of the time we're in production?

It is easy enough to have these boxes disappear completely, or be replaced by something else when we go into production. Pretty much a one line change to the code. We'll try to maximize the amount of interesting stuff in your screen real estate.

In both cases, you can see how the native isn't aligned correctly, this is most noticeable on the first. I'd say that you'd want to rotate it about 180 degrees about the x axis. Also, both of these show the scaling problem: in the relax phase, the RMSD hardly changes, so it might make sense to zoom that in, while on the second, the enegry has gone off the scale in a couple of places, making a zoom out make sense.

Having a one-size-fits all solution to the energy scaling is tricky because the ranges can be very large. Sometimes there will be a large spike in energy, and we don't want to rescale to that because all other detail would vanish. So very high energies just go off the chart. But I like your interactive zooming idea.

If somebody has some web space that they are willing to give up, I have an AVI of the screen saver in action that is 6551Kb, so that just about everyone can have a look, as I'm notoriously lazy about checking my e-mail, post here :)

I better give the screen capture guys a plug as I'm not buying their software (I just don't have the need) you can get it from ACS Capture 2.11

edit:

What I should of included was:

I don't know if the software (ACS capture 2.11) puts trojans, malware, spyware etc.. on your computer.
I don't know if the software does something weird with the AVI that may cause your computer harm.

You know, general I'm not to blame if something bad happens, type stuff.

I just downloaded it cause it looked about what I needed, wasn't too big, and had a trial that didn't include a "watermark".

---

Yup, I've got some... send me the file to both aieronpeters [at] yahoo [dot] co [dot] uk and johnathontinsley [at] hotmail [dot] com, and I'll upload it tonight.

---

Ok, I've set up an account and folder in my webspace, ANYONE can use this to upload pictures to be used HERE ONLY. If you want me to add a site (ie. your forums) then please send me an e-mail (below). If not, my leach protection will redirect you to one of my favorite websites! (www.premier.org.uk heeeeheee!!! :D).

Ok, username is : "freehost@blue-world.co.uk" ,
pass freehost,
ftp: ftp.blue-world.co.uk/

Note, the limit I have set is 10mb's on the webspace. If more is needed, again, e-mail me.

---

If all is well, I just sent you the file Johnathon.

---