I don't know if this has been discussed before. I don't recall seeing it, but forgive me if it has.

How about some kind of summary logging output from the Rosetta application. Perhaps for each trajectory done it could list the WU name, protein, the rmsd, the energy and any other data that may be interesting. The output could be a text file with one line per trajectory in a comma (or some other character) separated format. Then it shouldn't be too hard for a user to write a program to parse the log file and generate hes/her own stats and graphs. (I run mostly on a Linux box and have perl experience so I'm thinking a perl script to parse the log file and produce graphs via the GD modules.)

I know there are more important things to do right now like fixing well known bugs, but perhaps something like this could be added to the "long list of things to do".

Charlie

---

-Charlie

I don't know if this has been discussed before. I don't recall seeing it, but forgive me if it has.

How about some kind of summary logging output from the Rosetta application. Perhaps for each trajectory done it could list the WU name, protein, the rmsd, the energy and any other data that may be interesting. The output could be a text file with one line per trajectory in a comma (or some other character) separated format. Then it shouldn't be too hard for a user to write a program to parse the log file and generate hes/her own stats and graphs. (I run mostly on a Linux box and have perl experience so I'm thinking a perl script to parse the log file and produce graphs via the GD modules.)

I know there are more important things to do right now like fixing well known bugs, but perhaps something like this could be added to the "long list of things to do".

Charlie

This is a good idea and Rosetta of course does produce such a summary table at the end of each run. (which is what we look at to quickly evaluate the results of in house runs). I think currently boinc deletes all output files from the users machine when a job is done; perhaps it is possible to have an option which prevents this deleting. (this has to be an option, though, because most users won't want extra files accumulating on their machines).

---

This is a good idea and Rosetta of course does produce such a summary table at the end of each run. (which is what we look at to quickly evaluate the results of in house runs). I think currently boinc deletes all output files from the users machine when a job is done; perhaps it is possible to have an option which prevents this deleting. (this has to be an option, though, because most users won't want extra files accumulating on their machines).

David,

Thanks for the reply. Just to be clear, I was not suggesting keeping the output files from a WU around. I was suggesting a separate single file. Each WU would append a line to this file for each completed trajectory. Each line would have very minimal data - WU name, protien name, reported rmsd for the trajectory, reported low energy for the trajectory. Maybe also a time and date field. Obviously this file would grow over time. Maybe it could be made optional for those who don't want it.

On my Linux box I sometimes run "tail -f stdout.txt" in the rosetta slots directory to watch what's happening. I'll have to pay a bit more attention to see if the info I'm interested in is in there (and obvious to me.) I may be able to pull it from there.

Charlie

---

-Charlie

This sounds like a good idea for a addon app.

---

Another suggestion, on the Rosetta server status page, https://boinc.bakerlab.org/rosetta/rah_status.php ,
I would like to the the last time xml data was posted for Boinc-stats or Boinc-synergy to pick up. A schedule of how often the xml is re-crunched would be nice too.

Thanks,

dag

---

dag
--Finding aliens is cool, but understanding the structure of proteins is useful.

...
I would like to the the last time xml data was posted for Boinc-stats or Boinc-synergy to pick up. ...

This is already available on the index page to the XML files. Usually all the files should have the same date/time (within 1 sec or so)


Edit - added:
Likewise for the XML update times of all (most?) other BOINC projects I look at /stats/ directory under the project url

It is a rather nice example of re-using other people's code - by default Apache gives you this page whenever you *don't* provide an index.html file in a directory (notice the Apache ident at the bottom)

So in order to provide this info all the BOINC coders had to do was to make sure they did nothing :-)

... A schedule of how often the xml is re-crunched ...

on most BOINC projects it is daily. At one time it was more often on this project, don't know the current situation, but follow the above link for 24hrs and you will get the idea

hope that helps
River~~

---

Thanks.

dag

---

dag
--Finding aliens is cool, but understanding the structure of proteins is useful.