FORUMRunning Busco on Cedar
sjossey asked 5 months ago

  • Hi,

    I am trying to run busco on Cedar. I loaded the module and am trying to run with the following command
    $ python -hpython: can\’t open file \’\’:
    [Errno 2] No such file or directory

    So I tried,  $ /cvmfs/ -h usage: python -i [SEQUENCE_FILE] -l [LINEAGE] -o [OUTPUT_NAME] -m [MODE] [OTHER OPTIONS]
    Welcome to BUSCO 3.0.2: the Benchmarking Universal Single-Copy Ortholog assessment tool.For more detailed usage information, please review the README file provided with this distribution and the BUSCO user guide. optional arguments:-i FASTA FILE, –in FASTA FILE…. …

    That shows the help page  Now then if I try to run busco I get the following error
    $ /cvmfs/ -i genome.fasta -o Assemb  -l mammalia_odb9  -m geno ERROR No section [busco] found in /cvmfs/ Please make sure both the file and this section exist, see userguide.

    Please help if anybody was able to run busco on Cedar/Graham.

    Thanks, Sushma

    5 Answers
    Best Answer
    sjossey answered 5 months ago

  • I copied the file config.ini.default from  /cvmfs/ into my scratch folder and renamed it config.ini 
    Now from that folder I tried to run again got a new error
    $ python /cvmfs/ -i /genome.fasta -o Assemb -l mammalia_odb9 -m geno

    Traceback (most recent call last):
    File “/cvmfs/”, line 26, in <module>
    from pipebricks.PipeLogger import PipeLogger
    ImportError: No module named pipebricks.PipeLogger

    Eloi Mercier Staff replied 5 months ago

    It looks like it’s looking for a specific python module.

    What version of python are you using?
    python –version

    Eloi Mercier Staff replied 5 months ago

    A user reported that using python 3 worked for him:

    Can you try with python3?

    sjossey replied 5 months ago

    I did not load any python as I only loaded what was required by the module
    $module load nixpkgs/16.09 gcc/5.4.0 openmpi/2.0.2 busco/3.0.2

    sjossey replied 5 months ago

    I tried again after loading python/3.7.0..the same error

    Eloi Mercier Staff replied 5 months ago

    Can you run BUSCO on the test data please: -i /cvmfs/ -o TEST -l /cvmfs/ -m genome

    This ran without issue once I configure everything properly (see my next comment).

    Eloi Mercier Staff replied 5 months ago

    There is some configuration required before running BUSCO. Here is what I did. Perhaps it’s not the most elegant solution but it worked for me.

    1. Create a local copy of Augustus. Otherwise, if we use the version on cvmfs it will complain that it cannot write in augustus/config (“Cannot write to Augustus config path”):
    cp -r /cvmfs/ YOUR_WORKING_DIR

    2. Configure your config file properly. This should work on Cedar:
    # BUSCO specific configuration
    # It overrides default values in code and dataset cfg, and is overridden by arguments in command line
    # Uncomment lines when appropriate
    # Input file
    ;in = ./sample_data/target.fa
    # Run name, used in output files and folder
    ;out = SAMPLE
    # Where to store the output directory
    ;out_path = ./sample_data
    # Path to the BUSCO dataset
    ;lineage_path = ./sample_data/example
    # Which mode to run (genome / protein / transcriptome)
    ;mode = genome
    # How many threads to use for multithreaded steps
    ;cpu = 1
    # Domain for augustus retraining, eukaryota or prokaryota
    # Do not change this unless you know exactly why !!!
    ;domain = eukaryota
    # Force rewrite if files already exist (True/False)
    ;force = False
    # Restart mode (True/False)
    ;restart = False
    # Blast e-value
    ;evalue = 1e-3
    # Species to use with augustus, for old datasets only
    ;species = fly
    # Augustus extra parameters
    # Use single quotes, like this: ‘–param1=1 –param2=2’
    ;augustus_parameters = ”
    # Tmp folder
    ;tmp_path = ./tmp/
    # How many candidate regions (contigs, scaffolds) to consider for each BUSCO
    ;limit = 3
    # Augustus long mode for retraining (True/False)
    ;long = False
    # Quiet mode (True/False)
    ;quiet = False
    # Debug logs (True/False), it needs Quiet to be False
    ;debug = True
    # tar gzip output files (True/False)
    ;gzip = False
    # Force single core for the tblastn step
    ;blast_single_core = True

    # path to tblastn
    path = /cvmfs/soft.mugqic/CentOS6/software/blast/ncbi-blast-2.3.0+/bin/

    # path to makeblastdb
    path = /cvmfs/soft.mugqic/CentOS6/software/blast/ncbi-blast-2.3.0+/bin/

    # path to augustus
    path = YOUR_WORKING_DIR/augustus/3.3/bin

    # path to augustus etraining
    path = YOUR_WORKING_DIR/augustus/3.3/bin/

    # path to augustus perl scripts, redeclare it for each new script
    path = YOUR_WORKING_DIR/augustus/3.3/scripts
    path = YOUR_WORKING_DIR/augustus/3.3/scripts
    path = YOUR_WORKING_DIR/augustus/3.3/scripts

    # path to HMMsearch executable
    path = /cvmfs/soft.mugqic/CentOS6/software/hmmer/hmmer-3.1b2/bin/

    # path to Rscript, if you wish to use the plot tool
    path = /cvmfs/soft.mugqic/CentOS6/software/R_Bioconductor/R_Bioconductor-3.4.2_3.6/bin/

    3. Set up the environment variables:
    export AUGUSTUS_CONFIG_PATH=YOUR_WORKING_DIR/augustus/config/

    You should then be able to run BUSCO: -i /cvmfs/ -o TEST -l /cvmfs/ -m genome

    sjossey replied 5 months ago

    Thank you very much it seems to work on the test ..only thing I changed was while setting environment
    export AUGUSTUS_CONFIG_PATH=YOUR_WORKING_DIR/augustus/3.3/config

    Eloi Mercier Staff replied 5 months ago

    I’m glad it works!


    Eloi Mercier Staff answered 5 months ago

  • Hi Sushma,
    could you show me what command you used to load the module please?

    sjossey answered 5 months ago

  • Hi Eloi,
    I loaded using
    $module load nixpkgs/16.09 gcc/5.4.0 openmpi/2.0.2 busco/3.0.2

    jhgalvez Staff answered 5 months ago

  • You need to create the config file, even if it is empty. 
    Since you don’t have write access on CVMFS, you need to create it on a directory that you have access to and then define the environmental variable BUSCO_CONFIG_FILE. Otherwise the program won’t work, according to the documentation
    “You can set the BUSCO_CONFIG_FILE
    environment variable to define a custom path (including the filename) to the config.ini file,
    useful for switching between configurations or in a multi-users environment.”

    Eloi Mercier Staff replied 5 months ago

    On top of that, to run BUSCO, you simply need to type:
    (after loading the module)

    sjossey replied 5 months ago

    I will try to do that.

    jrosner Staff answered 4 months ago

  • Thread continues in another post

    Continued from the question….RUNNING BUSCO ON CEDAR