giraffe-parameter-search

Workflow for testing vg giraffe parameters

Contents:

• Experiment config file config.yaml
• Main snakefile Snakefile
• Sub-snakefile for graphing make_plots.smk
• Graphing util scatter.py
• Parameter search util parameter_search.py
• Parameter search util config file parameter_search_config.tsv

Note

for more information, visit the Wiki tab.

How To Use

1. Edit parameter_search_config.tsv to contain the parameters you want to generate values for, and what distribution you want to sample from. See parameter_search.py and the wiki for more information.

2. Run:

   python3 parameter_search.py --count x --benchmark-default
   

   This will create the hash_to_parameters file. You do NOT need to delete the old version of this file before running the script again. It will delete the file for you unless using --append.

3. Specify the statistics you want to measure and graphs you want to generate in config.yaml.

4. Run:

   snakemake --configfile config.yaml --config experiment=my_experiment --rerun-incomplete --executor slurm --keep-going
   

   Replace my_experiment with experiment of choice from config.

5. Your results will include a table of statistics, plots, and a manifest including the datetime, vg version, and config parameters.

Dependencies

Have the following:

• vg: Install vg here
• pandas
• yaml
• scipy
• bidict
• matplotlib

You can run:

conda create -n giraffe-param-search -c conda-forge -c bioconda snakemake=9.13.7 \
    snakemake-executor-plugin-slurm \
    bidict matplotlib pandas pyyaml scipy

TODO

• Add mismapped statistic
• Add variant calling sub snakefile