In Snakefile a set of rules are supplied on the basis of which output files are
supposed to be produced by the workflow.
It is customary to start with rule all a blank rule that uses all final output files as input files.
snakemake will go through the rest of the rules and create an execution sequence for all rules based
on the first rule. It will also determine which steps can be executed in parallel.
docker run -it --rm -v "$PWD":/app erblast/r_conda_snakemake_pkgs
snakemakesnakemake -n
snakemake -R `snakemake --list-code-changes`snakemake -Fsnakemake --cores 3The conda environment will be reconstructed from yml file and stored in ./.snakemake/conda.
A single conda environment can be defined for each rule.
conda env export --name snakemake_minimal -f ./envs/snake_minimal_macos.ymlsnakemake --use-condasnakemake -R `snakemake --list-code-changes` --use-conda --cores 3snakemake --dag | dot -Tpng > ./docs/wflow.pngsnakemake --report docs/index.html
config.yml
Scripts in R and python have access to a snakemake object carrying all rule
parameters as attributes. However when shell commands can be constructed snakemake's
parallel processing and logging capabilities can be leveraged.
R scripts can be added as .R or as .Rmd. When they are added as .Rmd they
can only produce one single html-output file. A workaround is to use an intermediate
R script as shown in rule.
see rules plot_rmd_direct and plot_rmd_via_script in Snakefile
Python scripts can be added as .py files. We can use papermill to execute
parametrized jupyter notebooks which we can then render as html. html is preferred
to notebooks because there is no doubt about the execution state.
see rules plot_execute_nb and plot_nb_2_html Snakefile
** the rules for rendering notebooks are not compatible with nb_conda as is.**
All common R functions are collected in an R package under utilR which is checked and tested
Execution times of each rule are stored in ./benchmark. Can be defined in Snakefile
unfortunately logging is not supported for scripts thus needs to be setup for each script individually using script-language-specific tools. https://bitbucket.org/snakemake/snakemake/issues/917/enable-stdout-and-stderr-redirection