Skip to content

Latest commit

 

History

History
143 lines (119 loc) · 5.89 KB

README.md

File metadata and controls

143 lines (119 loc) · 5.89 KB
Raw

snakemake

This repository works as a template to start a Snakemake project. Along with a basic file structure and example codes, it contains a snakemake profile to submit jobs on MARCC.

How to install Snakemake on MARCC

##########################################################
### load required modules                              ###
##########################################################

# load anaconda module version >= 4.6.0
module load  anaconda

# load python
module load python/3.7.4-anaconda 

############################################################
### create and activate a custom conda environment       ###
### to install/update packages without admin privilege   ###
### following instructions from                          ###
### https://www.marcc.jhu.edu/python-environments/.      ###
### see section "Case B. Custom conda environments"      ###
############################################################

# go to a directory to create conda 
# NOTE: MARCC recommends creating conda environments inside ~/work/code/
cd /home-1/[email protected]/python_env/conda # remember to change the directory

# create reqs.yaml file with basic packages
printf "dependencies:\n\
  - python=3.7\n\
  - matplotlib\n\
  - scipy\n\
  - numpy\n\
  - nb_conda_kernels\n\
  - au-eoed::gnu-parallel\n\
  - h5py\n\
  - pip\n\
  - pip:\n\
    - sphinx" > reqs.yaml
    
# install conda environment
conda env update --file reqs.yaml -p ./my_conda_env

# activate conda environment
conda activate /home-net/home-1/[email protected]/python_env/conda/my_conda_env

##############################################################################
### install snakemake using the new enviroment following instructions from ###
### https://snakemake.readthedocs.io/en/stable/tutorial/tutorial.html      ###
##############################################################################

# install mamba
conda install -c conda-forge mamba

# install snakemake using mamba
mamba create -c conda-forge -c bioconda -n snakemake snakemake

# exit from current environment
conda deactivate

# activate snakemake 
conda activate snakemake

# now you may run snakemake commands. test if help works.
snakemake --help

How to activate (and deactivate) Snakemake

module load  anaconda              # >= v4.6.0, use the version used during installation
module load python/3.7.4-anaconda  # >= v3.7, use the version used during installation
conda env list                     # you'll see all available environments
conda activate YOUR/SNAKEMAKE/ENV  # activate snakemake env
# conda deactivate                 # to exit/deactivate snakemake

How to run Snakemake on MARCC

  1. In your github account, create a new repository by using this repository as a template. Related tutorial. You will keep your codes in this repository. No need to write any extra code for demo.
  2. Clone the new repository on marcc. Related tutorial. 
    git clone https://github.com/USER-NAME/REPOSITORY-NAME
  3. Make sure you are on a MARCC login node and Snakemake is activated. See the section above.
  4. Go to the repository directory on MARCC. 
    cd PATH/TO/YOUR/REPO
  5. Run the following using maximum 2 cores. 
    snakemake --profile profiles/marcc -j2

How to configure jobs

Edit the global and job-specific configuration files to configure your jobs.

Global configuration file: profiles/marcc/config.yaml.

restart-times: 0                  # if failed, the job will not be restarted
jobscript: "slurm-jobscript.sh"
cluster: "slurm-submit.py"
cluster-status: "slurm-status.py"
max-jobs-per-second: 1            # max job submission rate 1 job/sec
max-status-checks-per-second: 10
local-cores: 1
latency-wait: 60                  # wait time (sec) if output file not found

Job-specific configuration file: profiles/marcc/cluster_config.yaml.

# default configuration for every rule (unless overridden)
__default__:
  partition: express
  nodes: 1
  ntasks: 1
  time: 10  # min
  output: "output/marcc_logs/{rule}/slurm-%j.out"
  error: "output/marcc_logs/{rule}/slurm-%j.err"
  job-name: "{rule}"

# configuration for "project_counts" rule -- overrides the default
project_counts:
  time: 15
  ntasks: 2

How to customize the repository for a specific project

  1. Add your scripts in the repository -- preferably in src folder (please create the folder).
  2. Edit configuration variables in config/config.yaml file.
  3. Add your Snakemake rules in rules folder.
  4. Edit Snakefile to aggregate all rules and to define final outcomes of the project.
  5. Edit profiles/marcc/cluster_config.yaml to allocate resources for each job.

NOTE: You may like to delete example rules in rules folder, example data in data/example folder.

Snakemake-related resources

  • Snakemake tutorial - Highly recommended!
  • Useful arguments to run Snakemake are available here.
  • Open-source snakemake profiles to run jobs on different environments are available here.
  • Snakemake-intro slides.
  • Snakemake in action: Ashis' project.
  • Snakemake video tutorial:Youtube link
  • How to deal with variable output (an unkown number of files) via checkpoints: Stack Overflow