ITG Unix Support
>    
     |  List directory  |  History  |  Similar  |  Print version  

HPC > IonMan Cluster > Programs > runjob

runjob

Description

runjob is the primary script around which the protein processing workflow on IonMan is based. Given a .trc, .inf, or .sym file, it processes a sample as far as it can go. When it can, it schedules work to be run simultaneously on other nodes using Condor

Synopsis

runjob <filename.inf>
runjob <filename.trc>
runjob <filename.sym>

Input

runjob reads the following files:

bin100.txt
pp3d.inp
go3d.inp
parent.txt
drftcal.inp
mak2.inp
mak3.inp

Output

runjob writes to the following files:

pp3d.inp (through rewrite-config)
go3d.inp (through rewrite-config)
perf.txt

Full Description

runjob takes a sample that you give it, and runs through the following programs, recording performance information into perf.txt:

  • trc300a? - loads TRC from /jobs, writes to serial scratch
  • bin100? - runs entirely in serial scratch
  • merge? - reads from serial scratch, writes to /jobs
  • c-pp3d? - schedules a pp3d job onto another node
  • (go3d)) - reads & writes to /jobs
  • drift_calib? - calibrates data
  • make_dta? - makes DTA's for charge state 2
  • c-makedta? - schedules a condor job to make DTA's for charge state 3
  • c-mascotsearch? - schedules a mascotsearch job for dta's 2 and 3

One feature of runjob is that you can resume a failed sample run without starting over. Here's how you do it.

  1. Fix whatever input files you need to get the sample to run again
  2. YOU MUST REMOVE condor/joblog.txt - this is the file that records the success or failure of your job. If it is there, it will move the job back to output or rejects without running anything.
  3. Re-run runjob

If you want runjob to re-run a particular program, you must remove the output files for that program. If re-running this program will change the results of programs that run afterwards, you must delete the results of those programs as well.

For instance, if you want to run go3d, you must remove it's peak file. Since the peak file is used to run drift_calib, which in turn is used to generate make_dta and mascot results, you must delete the output of those programs too.

These are the files you must remove if you would like a section to be re-run:

PROGRAM Filename it looks for
trc300a LC3D/LC3D1.d01
binpak LC3D/lc3d.a<binfactor>
merge SAMPLE.fra
go3d name of file in go3d.inp + .peak
pp3d name of file in pp3d.inp + .peak
drift_cal name of file in drftcal.inp + .abc
make_dta DTA- folders
mascotSearch ms-dtaname folders

Example

runjob YEAS3.trc

Relationships

runjob is called by jobhunter

runjob calls:

as well as the following scripts:

Exit Codes

21 Could not write to trc, bin100, or merge config file
22 Could not read a key/pair type configuration file
30 Could not determine parent/cid, incomplete frank files
31 drift_calib failed to generate output
38 go3d/pp3d completed, but a peak file is missing
39 go3d/pp3d completed, but a peak file is only 0 bytes
58 DTA jobs are complete, but dtalist.txt is missing.
59 DTA jobs are complete, but dtalist.txt is only 0 bytes
68 MascotSearch ran, but I raw file could not be found
69 MascotSearch ran, but I raw file is only 0 bytes
99 Could not schedule a task with Condor
* if a system() call exits with a code higher than 0, we exit with the same code.

Repository URL

http://revision.chem.indiana.edu/repos/ionman/scripts/runjob.rb

Author

Thomas Stromberg <tstrombe%indiana.edu>

 

Reference http://wiki.chem.indiana.edu/HPC/Runjob
Rights rw-rw-r--   tstrombe   IonMan

Prev. Programs used in the job processing workflow on IonMan   jobhunter Next