[phenixbb] Rosetta-MR - How long does it take?

Terwilliger, Thomas C terwilliger at lanl.gov
Fri Jan 27 10:54:41 PST 2012

Hi Tri,

It can take a long time to run mr_rosetta, but usually the slowest step is not  the molecular replacement step as in your case. Have a look at the directory where the work is being done to make sure that something is actually happening there.  At the end of your log file will be the file name of the log file for the current sub-process. At the end of that log file find the next one...etc.... Continue until you find the log file where the work is being done (see example below), and it should list an automr log file. This one will have the current Phaser output and you can see if it is doing anything.

The "place_model" step is normally running Phaser molecular replacement on however many input models from HHPRED you have specified.  If there are many NCS copies, or if one or more of these input models is very poor, then sometimes there may be Phaser runs that are never going to really work and that take a really long time.  If this is happening, then you might be better off to split your job into one run per search model. Then you can run them separately, toss any that take a really long time, and choose the one with the best final results.

All the best,
Tom T

For example:

Starting sub-processes Place model...

Splitting work into 2 jobs and running with 5 processors using sh
background=True in /net/marbles/raid1/terwill/misc/junk68/wt002/test_place_model/MR_ROSETTA_2/GROUP_OF_PLACE_MODEL_1

Starting job 1...Log will be: /net/marbles/raid1/terwill/misc/junk68/wt002/test_place_model/MR_ROSETTA_2/GROUP_OF_PLACE_MODEL_1/RUN_FILE_1.log

then looking in this log file:

Copies of search model to find: 1

Expected log for automr: /net/marbles/raid1/terwill/misc/junk68/wt002/test_place_model/MR_ROSETTA_2/GROUP_OF_PLACE_MODEL_1/RUN_1/AutoMR_run_1_/AutoMR_run_1_1.log

then look in the automr log file...if it is in the middle then it is doing something...
From: phenixbb-bounces at phenix-online.org [phenixbb-bounces at phenix-online.org] on behalf of Tri Ngo [ngoductri at gmail.com]
Sent: Friday, January 27, 2012 1:15 AM
To: phenixbb at phenix-online.org
Subject: [phenixbb] Rosetta-MR - How long does it take?

Hi Phenix users,

I tried to utilize the Rosetta-MR to solve an Esterase structure  (389 residues, 2 molecules/asu) which has 22% sequence identity to the know structure. I used HHPred to prepare the alignment input and Robetta fragment server to generate the fragment files. I also performed a quick test of mr-rosetta by running the command phenix_regression.wizards.test_command_line_rosetta_quick_tests. I think there is no error on my input files or the system.

However, the program never finished. It still shows me the "Placing Model..." status. It's running around 48 hours and still show the same log file (at "Running group of place_model" step). I am not sure if it's normal to wait days or weeks to get the work done when running Rosetta-MR? Thank you for your helpful input.

Best wishes,

Structural Biology Lab
School of Medicine - Sungkyunkwan University
Phone: 031-299-6150
Cell: 010-7774-8210

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20120127/3c1f81f8/attachment.htm>

More information about the phenixbb mailing list