The file config.yaml contains the configuration:
gaps: "data/hg38.gaps.bed"
genome: "/data/genomes/hg38/hg38.fa"
maxpeaks: 100
peak_dir: "data/remap.test"
pfm_dir: "data/pfm.test"
reference: "gimme.vertebrate.v5.0"
You will need to have the hg38 genome FASTA file available (shameless plug: use genomepy).
The peak_dir is set to data/remap.test and the pfm_dir to data/pfm.test. These are small data sets to check this to see if the workflow works. The variable maxpeaks selects the number of peaks to use (we used 5000 in the manuscript).
Use the script scripts/download_remap_peaks.sh to download all the remap peaks to the directory data/remap.
By setting pfm_dir to data/pfm all motif files named *.pfm in that directory will be included for comparison.
The reference determines which motif database will be used as a reference for figure1a.png.
When the workflow is finished, the directory out/ will contain several
final.*.txt files that contain all the metrics.
Install snakemake and gimmemotifs using conda:
conda create -n db_comparison python=3 snakemake gimmemotifs
Activate the environment:
conda activate db_comparison
Dry run:
snakemake -n
Full run:
snakemake --use-conda -j 12 --resources mem_mb=12000
Change the -j 12 to your preferred number of cores and --resources mem_mb=12000 to change the available memory (in MBs).
See cluster.json and the script submit_snakemake.sh for an example on how to
run the workflow on a cluster (SLURM in this case).