From Alliance Doc
Jump to navigation Jump to search
Other languages:

BUSCO (Benchmarking sets of Universal Single-Copy Orthologs) is an application for assessing genome assembly and annotation completeness.

For more information, see the user manual.

Available versions

Recent versions are available as wheels. Older versions are available as a module; please see the Modules section below.

To see the latest available version, run

[name@server ~]$ avail_wheel busco

Python wheel


1. Load the necessary modules.

[name@server ~]$ module load StdEnv/2020 gcc/9.3.0 python/3.10 augustus/3.5.0 hmmer/3.3.2 blast+/2.13.0 metaeuk/6 prodigal/2.6.3 r/4.3.1 bbmap/38.86

2. Create the virtual environment.

[name@server ~]$ virtualenv ~/busco_env
[name@server ~]$ source ~/busco_env/bin/activate

3. Install the wheel and its dependencies.

(busco_env) $ pip install --no-index biopython==1.81 pandas==2.1.0 busco==5.5.0

4. Validate the installation.

(busco_env) $ busco --help

5. Freeze the environment and requirements set. To use the requirements text file, see the bash submission script shown at point 8.

(busco_env) $ pip freeze > ~/busco-requirements.txt



6. You must pre-download any datasets from BUSCO data before submitting your job.

You can access the available datasets in your terminal by typing busco --list-datasets.

You have two options to download datasets:

  • use the busco command,
  • use the wget command.
6.1 Using the busco command

This is the preferred option. Type this command in your working directory to download a particular dataset, for example

[name@server ~]$ busco --download bacteria_odb10

It is also possible to do a bulk download by replacing the dataset name by the following arguments: all, prokaryota, eukaryota, or virus, for example

[name@server ~]$ busco --download virus

This will

1. create a BUSCO directory hierarchy for the datasets,
2. download the appropriate datasets,
3. decompress the file(s),
4. if you download multiple files, they will all be automatically added to the lineages directory.

The hierarchy will look like this:

  • busco_downloads/
  • information/
  • lineages/
  • placement_files/

Doing so, all your lineage files should be in busco_downloads/lineages/. When referring to --download_path busco_downloads/ in the BUSCO command line, it will know where to find the lineage dataset argument --lineage_dataset bacteria_odb10. If the busco_download directory is not in your working directory, you will need to provide the full path.

6.2 Using the wget command

All files must be decompressed with tar -xvf file.tar.gz.

[name@server ~]$ mkdir -p busco_downloads/lineages
[name@server ~]$ cd busco_downloads/lineages
[name@server ~]$ wget https://busco-data.ezlab.org/v5/data/lineages/bacteria_odb10.2020-03-06.tar.gz
[name@server ~]$ tar -xvf bacteria_odb10.2020-03-06.tar.gz


7. Download a genome file.

[name@server ~]$ wget https://gitlab.com/ezlab/busco/-/raw/master/test_data/bacteria/genome.fna

8. Run.

Command to run a single genome:

[name@server ~]$ busco --offline --in genome.fna --out TEST --lineage_dataset bacteria_odb10 --mode genome --cpu ${SLURM_CPUS_PER_TASK:-1} --download_path busco_download/

Command to run multiple genomes that would be saved in the genome directory (in this example, the genome/ folder would need to be in the current directory; otherwise, you need to provide the full path):

[name@server ~]$ busco --offline --in genome/ --out TEST --lineage_dataset bacteria_odb10 --mode genome --cpu ${SLURM_CPUS_PER_TASK:-1} --download_path busco_download/

The single genome command should take less than 60 seconds to complete. Production runs which take longer must be submitted to the scheduler.

BUSCO tips

Specify --in genome.fna for single file analysis.

Specify --in genome/ for multiple files analysis.

Slurm tips

Specify --offline to avoid using the internet.

Specify --cpu to $SLURM_CPUS_PER_TASK in your job submission script to use the number of CPUs allocated.

Specify --restart to restart from a partial run.

Job submission

Here you have an example of a submission script. You can submit as so: sbatch run_busco.sh.

File : run_busco.sh


#SBATCH --job-name=busco9_run
#SBATCH --account=def-someprof    # adjust this to match the accounting group you are using to submit jobs
#SBATCH --time=01:00:00           # adjust this to match the walltime of your job
#SBATCH --cpus-per-task=8         # adjust depending on the size of the genome(s)/protein(s)/transcriptome(s)
#SBATCH --mem=20G                 # adjust this according to the memory you need

# Load modules dependencies.
module load StdEnv/2020 gcc/9.3.0 python/3.10 augustus/3.5.0 hmmer/3.3.2 blast+/2.13.0 metaeuk/6 prodigal/2.6.3 r/4.3.1 bbmap/38.86

# Generate your virtual environment in $SLURM_TMPDIR.
virtualenv --no-download ${SLURM_TMPDIR}/env
source ${SLURM_TMPDIR}/env/bin/activate

# Install busco and its dependencies.
pip install --no-index --upgrade pip
pip install --no-index --requirement ~/busco-requirements.txt

# Edit with the proper arguments, run your commands.
busco --offline --in genome.fna --out TEST --lineage_dataset bacteria_odb10 --mode genome --cpu ${SLURM_CPUS_PER_TASK:-1} --download_path busco_download/

Augustus parameters

9. Advanced users may want to use Augustus parameters: --augustus_parameters="--yourAugustusParameter".

  • Copy the Augustus config directory to a writable location.
[name@server ~]$ cp -r $EBROOTAUGUSTUS/config $HOME/augustus_config
  • Make sure to define the AUGUSTUS_CONFIG_PATH environment variable.
[name@server ~]$ export AUGUSTUS_CONFIG_PATH=$HOME/augustus_config

SEPP parameters

10. To use SEPP parameters, you need to install SEPP locally in your virtual environment. This should be done from the login node.

10.1. Activate your BUSCO virtual environment.

[name@server ~]$ source busco_env/bin/activate

10.2. Install DendroPy.

[name@server ~]$ pip install 'dendropy<4.6'

10.3. Install SEPP.

[name@server ~]$ git clone https://github.com/smirarab/sepp.git
[name@server ~]$ cd sepp
[name@server ~]$ python setup.py config
[name@server ~]$ python setup.py install

10.4. Validate the installation.

[name@server ~]$ cd
[name@server ~]$ run_sepp.py -h

10.5. Because SEPP is installed locally, you cannot create the virtual environment as described in the previous submission script. To activate your local virtual environment, simply add the following command immediately under the line to load the module:

[name@server ~]$ source ~/busco_env/bin/activate



This section is outdated and deprecated. You should use the wheels available.

1. Load the necessary modules.

[name@server ~]$ module load StdEnv/2018.3 gcc/7.3.0 openmpi/3.1.4 busco/3.0.2 r/4.0.2

This will also load modules for Augustus, BLAST+, HMMER and some other software packages that BUSCO relies upon.

2. Copy the configuration file.

[name@server ~]$ cp -v $EBROOTBUSCO/config/config.ini.default $HOME/busco_config.ini


[name@server ~]$ wget -O $HOME/busco_config.ini https://gitlab.com/ezlab/busco/raw/master/config/config.ini.default

3. Edit the configuration file. The locations of external tools are all specified in the last section, which is shown below:

File : partial_busco_config.ini

# path to tblastn
path = /cvmfs/soft.computecanada.ca/easybuild/software/2017/avx512/Compiler/gcc7.3/blast+/2.7.1/bin/

# path to makeblastdb
path = /cvmfs/soft.computecanada.ca/easybuild/software/2017/avx512/Compiler/gcc7.3/blast+/2.7.1/bin/

# path to augustus
path = /cvmfs/soft.computecanada.ca/easybuild/software/2017/avx512/Compiler/gcc7.3/augustus/3.3/bin/

# path to augustus etraining
path = /cvmfs/soft.computecanada.ca/easybuild/software/2017/avx512/Compiler/gcc7.3/augustus/3.3/bin/

# path to augustus perl scripts, redeclare it for each new script
path = /cvmfs/soft.computecanada.ca/easybuild/software/2017/avx512/Compiler/gcc7.3/augustus/3.3/scripts/
path = /cvmfs/soft.computecanada.ca/easybuild/software/2017/avx512/Compiler/gcc7.3/augustus/3.3/scripts/
path = /cvmfs/soft.computecanada.ca/easybuild/software/2017/avx512/Compiler/gcc7.3/augustus/3.3/scripts/

# path to HMMsearch executable
path = /cvmfs/soft.computecanada.ca/easybuild/software/2017/avx512/Compiler/gcc7.3/hmmer/3.1b2/bin/

# path to Rscript, if you wish to use the plot tool
path = /cvmfs/soft.computecanada.ca/easybuild/software/2017/avx512/Compiler/gcc7.3/r/4.0.2/bin/

4. Copy the Augustus config directory to a writable location.

[name@server ~]$ cp -r $EBROOTAUGUSTUS/config $HOME/augustus_config

5. Check that it runs.

[name@server ~]$ export BUSCO_CONFIG_FILE=$HOME/busco_config.ini
[name@server ~]$ export AUGUSTUS_CONFIG_PATH=$HOME/augustus_config
[name@server ~]$ run_BUSCO.py --in $EBROOTBUSCO/sample_data/target.fa --out TEST --lineage_path $EBROOTBUSCO/sample_data/example --mode genome

The run_BUSCO.py command should take less than 60 seconds to complete. Production runs which take longer should be submitted to the scheduler.


Cannot write to Augustus config path

Make sure you have copied the config directory to a writable location and exported the AUGUSTUS_CONFIG_PATH variable.