This page last changed on Mar 06, 2007 by maising.
  • Install the pipeline prerequisites on a suitable Linux system and compile the pipeline
    make
    make install
    

    See Pipeline installation for details.

  • If you have access to the pipeline validation DVD, run the validation script "validate.sh" on the test data. See the documentation that comes with the DVD for details.
  • For the first run of the pipeline for a new instrument, make sure that you have set up the directory "Instruments" for calibration files as described in Pipeline usage.
  • Copy the run data (which sits in a folder structure called the run-folder) to a location that is visible to the analysis pipeline and your analysis computing system.
  • If you want to do sequence alignment you need to create a config file telling the pipeline what type of analysis against which targets to do for each lane. The format and contents of the config file ("config.txt") is described in GERALD User Guide and FAQ.
  • Change into the run-folder directory (a directory of the form "YYMMDD_INSTRUMENT_XXXX_FCYYYY"):
    cd run-folder
    

    The usage of the pipeline is explained in more detail in Pipeline usage.

  • Generate Makefiles:
    /path-to-Pipeline/Goat/goat_pipeline.py . --offsets=auto --make --GERALD=path/config.txt
    

    Command-line options are explained in more detail in GOAT documentation - Firecrest, Bustard

  • Change into the newly created folder "Data/C1...Firecrest..."
  • Start the analysis run through make, possibly using the parallelisation switch "-j" with a setting appropriate to your analysis computing system:
    nohup make -j 4 recursive
    

    More information on parallelisation can be found in Pipeline parallelisation.

Document generated by Confluence on Apr 05, 2007 11:25