I used conda and bioconda to manage software dependencies. To replicate the computing environment, you will need to complete the following 4 steps. Note that this is only guaranteed to work on a Linux-64 based architecture.
- Download and install Miniconda (instructions)
- Download environment file. If you cloned the Git repo, you already have the environment file.
conda install anaconda-client; anaconda download jdblischak/cardioqtl
- Create conda environment:
conda env create -n cardioqtl --file environment.yaml
- Activate conda environment:
- To activate:
source activate cardioqtl - To deactivate:
source deactivate cardioqtl
- To activate:
The code is freely available for reuse with attribution via the MIT license.
-
Snakefile- Implements the analysis pipeline -
submit-snakemake.sh- Submits individual jobs produced inSnakefileto Slurm. If your cluster uses a different job scheduler, you'll need to edit this file andcluster.json. -
scripts/- R scripts called bySnakefile -
scratch/- Exploratory analyses written in R Markdown
-
data/counts-subread.txt- Gene counts after mapping to GRCh37 with Subjunc and summing counts per gene with featureCounts (Subread 1.5.0p3). Includes all genes in Ensembl release 75 (i.e. protein coding plus all other biotypes; seescripts/create-exons.R). -
data/counts-clean.txt- Gene counts after removing samples 26302, 110232, and 160001 and removing genes with log2 cpm less than 0 (seescripts/clean-counts.R). -
data/counts-normalized.txt- Gene counts after normalizing to N(0,1) within each sample follwed by normalizing to N(0,1) within each gene. Used the R functionqqnorm(seescripts/normalize-counts.R).