Posts by Christoph Jansen

1) Message boards : Cafe Rosetta : New Protein Folding Project on BOINC...POEM@home (Message 47830)
Posted 18 Oct 2007 by Profile Christoph Jansen
Just forget it, I am out of here.
2) Message boards : Number crunching : Problems with Rosetta version 5.80 (Message 47692)
Posted 13 Oct 2007 by Profile Christoph Jansen
The machines affected by this "multi-decoy" bug seem all to be Core2Quads. And, contrary to what the 4000 credit WU thread implies, also other WU types than mcr1
are affected, see these two:
3) Questions and Answers : Windows : why GPU can not work this project!!!!! (Message 47652)
Posted 12 Oct 2007 by Profile Christoph Jansen
[First: Folding at home uses GPUs, but the client is far from easy to install. But you probably know that already.]

On the topic itself: As far as I know one of the biggest problems standing against running Rosetta on GPUs is that Rosetta is developing so quick. Each client has to be programmed and beta tested to ensure there are as little problems as possible. Both takes its time.

This also the reason why Rosetta does not only have no GPU clients, but also no clients optimised for Intel and AMD CPUs separately, which many other projects have. The more clients you produce the more development time has to be spent on coding and testing.

And as - as I have said - the people at Rosetta enhance their client pretty often and have very limited human resources for programming and testing they stick to the number and kind of clients that are currently available.

Unless the Bakerlab people do not get access to additional resources or get the rosetta project into a phase where enhancements become slower because they are kind of nearing a "perfect" folding algorithm, this will probable remain the state of things.
4) Message boards : Rosetta@home Science : DISCUSSION of Rosetta@home Journal (3) (Message 47592)
Posted 10 Oct 2007 by Profile Christoph Jansen
Something that fits exactly in here: I just heard that Gerhard Ertl receives the 2007 Nobel Prize for Chemistry. He is awarded the prize for his studies on understanding the mechanisms of chemical catalysis. Literally, the Nobel Price Committee awarded the price "for his studies of chemical processes on solid surfaces".

Industrial scale process catalysis is mostly achieved by having a solid state catalyst on which gases or liquids can react. Take your catalytic converter in your car as an example. Understanding catalysis in terms of a chain of

adsorbtion of reactants => reaction => desorbtion of products

has long been very hard to do, because it is a thing that happens on the atomic scale und is very hard to observe. So what you basically do is collect a lot of data measured on your catalyst/reactant system and try to interpret these in terms of what happens on the interface between both. By understanding known catalytic reactions researchers hope to refine important processes, but also to find possible new ones.

Now move your imagination away from this kind of "groping in the dark" approach. Imagine you would not only be able to understand cataylsis on the molecular scale, but that you can design it. Design it because you can exactly calculate which protein structure you need to synthetise a desired molecule from given reactants.

And then all you need is to produce the DNA or RNA strand that codes this protein and let bacteria or other cells produce it. As I already said in my other post, this is taking catalysis research far beyond what scientists thought they could ever do in their wildest dreams.
5) Message boards : Rosetta@home Science : DISCUSSION of Rosetta@home Journal (3) (Message 47573)
Posted 9 Oct 2007 by Profile Christoph Jansen
[...] the design of novel enzymes (using computational protein design and rosetta@home) that catalyze chemcial reactions for which there are no naturally ocurring catalysts. [...] This opens the door to creating new catalysts for all sorts of chemical reactions that are difficult currently or require toxic or environmentally unfriendly reaction conditions.

This is exceptional in two respects.

First: It has long been a dream of chemists to be able to catalyse chemical reactions in the same efficient way nature does so using enzymes. One example that illustrates it quite impressingly is the issue of nitrogen fixation. Ammonia is one of the most important raw materials in the chemical industries. It is used directly as well as indirectly, the latter mostly a nitrates, in fertilisers or as one of the major raw materials for ammunition and explosives.

If you want to make ammonia from nitrogen and hydrogen you need the so called "Haber Bosch Process", which basically means you have to heat both to 450°C (840°F) and 300 bars (4500 psi) and bring them in contact with a special catalyst. This process proved so important for agriculture (and alas also for Germany's capability of leading World War I), that Haber and Bosch got the Nobel Prize for it. A lot of plants on the other hand synthetise ammonia at ambient temperature.

Building a catalyst that can rival with what plants do (or any other life form does) would mean saving a lot of energy and thus money (and today also has a lot of value in saving the emission of lots of greenhouse gases).

Now, attempts to copy the "active sites" of proteins - the places that do the reactions - have long been made, with more or less success. If you have noticed that today you can do laundry at 40°C, well, it is due to one of these attempts. This resarch procuced a catalyst that activates the peroxy-compounds doing the bleaching and killing bacteria at considerably lower temperatures than before, when only heat could activate the release of oxygen from the peroxy compounds.

Still, many of the results achieved to date are pretty crude even if we know the exact structure of the proteins involved. So being able to produce enzymes that can do known processes very efficiently would already be a revolution in itself.

But second:

To go forth claiming to have designed a wholly new protein unmatched in nature which does a reaction that has not previously been catalysed by proteins goes right beyond this goal and out to the other side: It's a milestone on the way to true catalyst design for any reaction you like - and it has the potential to revolutionize the chemical industries.

There are so many implications that it is hard to even start describing them. Just think of processes in terms of the nitrogen fixation: faster, cleaner, safer, cheaper.
6) Message boards : Number crunching : Problems with Rosetta version 5.80 (Message 47316)
Posted 2 Oct 2007 by Profile Christoph Jansen
Now finally BOINC hung when I tried to shutdown my computer (I did't notice that for about an hour). This
resulted in all of the rest of the WUs still present on the drive to error out, whatever the cause for that may be.

BOINC probably finished a WU and tried to execute the next one but didn't get any system resources to do so as the shutdown
process was meant to go on. So one after one they all marched straight into oblivion. Poor wretches...
7) Message boards : Number crunching : Problems with Rosetta version 5.80 (Message 47295)
Posted 1 Oct 2007 by Profile Christoph Jansen
This one:

keeps on stopping, meaning the processor time keeps stopping to count up indefinitely, so the watchdog does not shut it down.

I had that on another one over last night, it must have been stuck for some hours. Didn't note the number though.
8) Message boards : Rosetta@home Science : What is the unit of "Accepted Energy?" (Message 46815)
Posted 22 Sep 2007 by Profile Christoph Jansen
Interesting they still work with kilocalories. So you need to multiply all values by roughly 4 (4,18605 exactly) to get kilojoules.
9) Message boards : Rosetta@home Science : What is the unit of "Accepted Energy?" (Message 46723)
Posted 21 Sep 2007 by Profile Christoph Jansen
I'd say it is kJ/mole (Kilojoules per mole, with mole meaning 6*10^23 Particles, not the animal...), which is the normal units assigned to molecular formation energies. The ranges I see in Rosetta go well with that:

1 mole of carbon produces around 400 kJ when burned to CO2.

Hydrogen bonds and other weak molecular interactions have some percents to per thousands of that. Most of the proteins we calculate have hundreds to thousands of such interactions, so they should also be roughly in a range of some tens to some hundreds of Kilojoules per mol. And this is exactly what I see most of the time.

[Edit because of unsuccessful attempts to get a proper subscript in CO2...)
10) Message boards : Number crunching : German: Boinc Error, tasks overdue (Message 46062)
Posted 12 Sep 2007 by Profile Christoph Jansen
Sieht ja alles ganz normal aus. Laufen denn jetzt noch andere Work Units, oder hast Du BOINC erstmal abgeschaltet? Wäre ja interessant, ob auch mal welche bis zum Ende durchhalten.

Ich habe leider überhaupt keine Erfahrung mit Vista, darum kann ich nichts dazu sagen. Vielleicht hat ja irgendjemand sonst hier im Forum Ahnung davon.

Ansonsten hatte ich auch Probleme mit einigen WUs der letzten Tage. Aber da war es immer so, dass sie einfach an einem bestimmten Punkt festhingen und dann wegen Zeitüberschreitung abgebrochen wurden.
11) Message boards : Number crunching : German: Boinc Error, tasks overdue (Message 46057)
Posted 12 Sep 2007 by Profile Christoph Jansen
Die Berechnungen brechen mit einem Fehler ab. Das kann eine ganze Menge von Ursachen haben, von Problemen mit dem Betriebssystem über zu hohe Temperatur oder Speicherfehler bis hin zu zu aggresivem Übertakten. Ist der Rechner übertaktet oder läuft er mit "stock speed"? Wie sind die Temperaturen?

For mods:

Your computations are aborted with an error. This can have a number of causes from issues with your operating system and high temperatures or memory failures to aggressive overclocking. Is your machine overclocked or does it run at stock speed? And what are your temperatures?
12) Message boards : Number crunching : German: Boinc Error, tasks overdue (Message 46003)
Posted 11 Sep 2007 by Profile Christoph Jansen
Mi Mod.Sense

a short overview:

pool says his machine claims to have an overdue work unit in its directory, which is strange because he has only attached to the project yesterday. His "results" for the attached machine are empty, so no WUs have been downloaded to that machine.

I asked him to try a reset and see if his machine will download new WUs after that.

Another guess I made is that he erroneously attached to another account than the one he is posting with which he can see by going to "Projects" in the BOINC manager, highlighting Rosetta@home and then going to the "Your Account" Button. If the same account he is posting with opens, the reset might have done it. If it does not, he's still got another account.
13) Message boards : Number crunching : German: Boinc Error, tasks overdue (Message 46002)
Posted 11 Sep 2007 by Profile Christoph Jansen
Hallo pool,

wenn Du mal auf Deinen Account gehst (diesen Link klicken) und dann "Computers on this Account" anklickst, siehst Du, dass dort ganz rechts unter "Results" eine Null steht. Dein Computer hat sich demnach noch überhaupt keine Work Units heruntergeladen. Oder hast Du zwei Accounts?

Was steht denn in Deinem BONC Manager unter dem Tabellenreiter "Aufgaben"? Sind da Work Units angegeben, die Dein Rechner gerade bearbeitet, oder sind die Aufgaben leer?

Eine relativ einfache Möglichkeit wäre, einfach im BOINC Manager auf den Reiter "Projekte" zu klicken, das Projekt "Rosetta@home" anzuklicken und dann links auf den Button "Zurücksetzen" zu klicken. Das entfernt alle derzeitigen Rosetta-Work Units aus Deinem BOINC-Verzeichnis. Berichte danach mal, ob der Rechner sich dann neue Arbeit herunterlädt.

Und außerdem solltest Du mal im BOINC Manager, ebenfalls unter "Projekte" auf den Button "Ihr Account" klicken und nachsehen, ob sich dann wirklich der Account öffnet, mit dem Du Deine Anfrage gepostet hast. Vielleicht hast Du beim Anmelden den Projekts einen Tippfehler drin und hast jetzt zwei Accounts? Das passiert schnell, wenn man ein Projekt anmeldet und sich nicht in einen vorhandenen Account einloggt (der Anmeldedialog gibt beides zur Auswahl, standardmäßig ist "Nein, neues Teilnehmerkonto" aktiviert).
14) Message boards : Number crunching : T-FLOPS jump?? (Message 45937)
Posted 10 Sep 2007 by Profile Christoph Jansen
Is it just me or has Rosetta T-FLOPS estimate jumped from ~50 before the crash to about 85 afterwards! could this be due to the influx of results once it came back online or have people re-doubled their efforts?

It's due to incoming results genearaed during the project's downtime.

The Teraflops estimate on the main page is not a long-time moving average like RAC but rather a daily value. It always correlates directly to the value given under "Credits last 24h". Right now it says:

Credits last 24h : 8,446,053

TeraFLOPS estimate: 84.461

In other words: Credits during last 24 hours divided by 100.000 gives you the current TFlops.
15) Message boards : Number crunching : Problems with Rosetta version 5.78 (Message 45936)
Posted 10 Sep 2007 by Profile Christoph Jansen
Same here too:

"Rosetta score is stuck or going too long. Watchdog is ending the run!"

On these WUs:

16) Message boards : Number crunching : My protein fell apart :( (Message 45201)
Posted 20 Aug 2007 by Profile Christoph Jansen
That special protein must be biodegradable, but that takes bacteria - so it seems like your machine is infected by something utterly new...

OK, w/o kidding: I noticed that myself some time ago, although not to such an extent. Looks some kind of Lego kit, but nothing to worry about.
17) Message boards : Number crunching : Claiming 30 cred and getting 9200 granted? (Message 45196)
Posted 20 Aug 2007 by Profile Christoph Jansen
Or maybe decoys are being completed extremely quickly, maybe a bug in the application? It may not even be generating valid results.

There are also some results with "computing error" in the user's list, so I'd say it is a matter of instability just at the brink of failure, whatever the reason (temperature, overclock, bad memory...).

I think this to be the most probable explanation beacuse it looks a little like the "just get it to calculate rubbish and finish quick" thing that buggered the classic SETI@home: Some people used to get their machines to finish WUs with invalid results in almost no time by overclocking them. Got them lots and lots of finished WUs a day without any valuable contribution (on the contrary: as there was no quorum it was rather destructive to the science).

But I would in no way imply it is done intentionally, mind. It is just the resemblance of the phenomenon. The user surely does not know his machine needs a little tending to at the moment.
18) Questions and Answers : Windows : System with Deepfreeze - How to save work units? (Message 44401)
Posted 27 Jul 2007 by Profile Christoph Jansen
Hi Zxian,

DeepFreeze does not affect a number of folders/drives depending on its settings. Maybe this part of the DeepFreeze documentation helps you:

Retaining User Data

Have a look over it an if questions come up on how-tos just ask again.


19) Message boards : Rosetta@home Science : Bioengineered protein cuts HIV genome and frees cells from infection (Message 44324)
Posted 26 Jul 2007 by Profile Christoph Jansen
So what's the catch? Is AIDS cured once and for all?

No, AIDS is of course not yet cured, as the protein used may not qualify as a drug candidate at all when tested. But the thing those researchers did with their artificial or promoted evolution is to create a protein that does a very specific task, and that is exactly what Rosetta is doing in terms of protein design. The difference being that Rosetta has the potential do it in much shorter time.

Creating proteins that cut out or disactive specific genes is one of the major goals in fighting diseases, as practically everything that is disease related in our body is finally caused or promoted by proteins. How these proteins can be administered as drugs depends on a number of things.

You can bring proteins into the body by infusions or similar if they are to be of general use, or put them on carrier substances if you want them to reach a specific target organ/cell type but do not want them to be present all the time. One example of such substances are vaccines, they are meant to create an immunity by contact with your immune system and are no longer needed after that. Another one are cancer treatments. Once the cancer is gone they are no longer needed either, maybe they are even harmful. And finally you can bring DNA strands into the body, that are inserted in cells temporarily, meaning not going into the nucleus. These strands can produce specific proteins, e.g. fighting certain viruses or bacteria, but will eventually be metabolized, i.e. decomposed, and stop producing the protein. Such a drug DNA strand can be administered like a typical medicine, just when it is needed.

And there is the possibility of inserting new genes into your genetic code permanently using special carriers. You can e.g. do that to cure hereditary diseases or acquired genetic defects, maybe also to fight diseases like AIDS, where the virus is very persistent and you may need to fight it again and again as small portions of it can reside in your body no matter how long the treatment is administered.

In all these cases current methods to identify the proteins involved and design the proteins/DNA that can address the genetic or proteomic causes are so called "wet bench" methods, that means they are extremely time and cost intensive.

But having proven that these slow and expensive methods can achieve a breakthrough in destroying the HIV virus in a cell culture only underlines the importance of Rosetta as Rosetta, once mature, gives researchers a tool to come to such results much, much quicker and more economically. In other words, they can create a lot more possible ways to eventually find that one cure for AIDS that also qualifies as a drug you can administer to humans, which is the real problem in any drug discovery process.
20) Message boards : Number crunching : Computers without internet (Message 44122)
Posted 23 Jul 2007 by Profile Christoph Jansen

first of all: why don't you just get yourself a PCI Etnernet card for the PC? Just stick it in a free slot and you'll have an Ethernet port again. These things just cost a few dollars.


Next 20

©2023 University of Washington