ENCODE4 distal regulation E-G correlation predictors

Compute simple enhancer-gene regulatory predictions based on correlation of chromatin accessibility at cCREs and accessibility at gene promoters or RNA expression levels of genes across different biosamples.

Computing DNase-DNase correlations

To compute pearson and spearman correlation coefficients and Generalized Least Square regression coefficients between DNase-seq signal at cCREs and DNase-seq signal at promoters within 1Mb, run the following commands. This also installs all dependencies using conda and downloads DNase-seq bam files from the ENCODE portal. Use the scratch parameter in the config.yml to specify where temporary bam files should be stored.

Because of the large number of correlations that are computed and the memory and disk space requirements, the current implementation of the workflow is only feasible to be executed on an HPC system.

# edit 'scratch' parameter to specify location to store bam files, e.g. scratch
vim config/config.yml

# compute all DNase-DNase correlation using parallelization (-n = dryrun, remove to run). adapt this
# command for submission to your cluster based on snakemake documentation
snakemake --use-conda results/DNase/correlation.DNase_seq-DNase_seq.100bp.tsv.gz -j100 -n

Computing DNase-RNA correlations

The workflow can also compute correlations between DNase-seq signal at enhancers with RNA expression of protein-coding genes with promoters within 1Mb. RNA expression is taken from ENCODE RNA-seq gene quantification files. To download these files and compute DNase-RNA correlations simply run the following commands:

# edit 'scratch' parameter to specify location to store bam files, e.g. scratch
vim config/config.yml

# download DNase-seq and RNA-seq data, and compute DNase-RNA correlations
snakemake --use-conda results/RNA/correlation.DNase_seq-RNA_seq.100bp.tsv.gz -j100 -n

Computing correlation metrics for other sample lists

To compute DNase-DNase or DNase-RNA correlations for other sets of samples, add ENCODE bam and gene quantification accession IDs to either config/dnase_dnase_corr_input_files.tsv or config/dnase_rna_corr_input_files.tsv. New sample sets can also be provided in new files following the same format, which can be added to the workflow in the config/config.yml. For DNase-seq bam files the workflow also requires information whether the reads are single- or paired-ended in config/dnase_dnase_corr_dnase_bam_metadata.tsv or dnase_rna_corr_dnase_bam_metadata.tsv, respectively in a new file following the same format.