README assumes that this repo will be cloned to ./bismsmark, and assumes that conda/miniconda has been installed.
git clone https://github.com/lyijin/bismsmark
cd bismsmark
conda create -c conda-forge -n bismsmark mamba
conda activate bismsmark
mamba install -c bioconda -c conda-forge snakemake bismark trim-galore
(skip next three commands if running locally. if running on cluster, with slurm as scheduler, run these commands)
mkdir -p ~/.config/snakemake/slurm
nano ./config/slurm/config.yaml and add your email, so that slurm can email you if things fail. Remove the last two flags if you don't want to be bothered / just want to test things out quickly.
cp ./config/slurm/config.yaml ~/.config/snakemake/slurm
cd workflow
(use this command if running locally.)
snakemake --cores 8
(use this command if running on slurm)
snakemake --profile slurm
cat Snakefile | head -40
Use the test files to guide naming choices.
To be run AFTER the snakemake pipeline completes successfully.
The Python script compile_bismark_logs.py scrapes some stats from bismark logs e.g. mapping rates, deduplication rates, ...
Run this script from workflow/, i.e. python3 scripts/compile_bismark_logs.py. No additional arguments needed, as the script hardcodes the folder names containing log files.
The resources (RAM/cores) requested in Snakefile is optimised to work on the human genome (hg38/GRch38), and on my local slurm cluster. Remember that larger genomes will need more RAM, larger sequence files will need more time to be processed.
To check resources consumed by jobs, use sacct with the -S mmdd flag, where mmdd == month, date e.g. -S 0801.
Erm, it's because I'm trying to embed snakemake ("sm") into bismark to make a reproducible pipeline.