Context-Adjusted Proportion of Singletons (CAPS)

Code for the paper

Tested with Hail version 0.2.107 and Snakemake 7.32

The "Data" pipeline

data/ contains Hail and Snakemake code that requires execution in Google Cloud and saves files to a Google Storage (GS) bucket. Copies of the generated files are available in files/.

How to run

1. Create a new cluster: hailctl dataproc start <cluster_name> --packages snakemake --requester-pays-allow-buckets gnomad-public-requester-pays --project <project_name> --bucket <bucket_name> --region <region> --num-workers <N> --image-version=2.0.27-debian10
2. Connect to the cluster: gcloud beta compute ssh <user_name>@<cluster_name>-m --project "<project_name>"
3. git clone this repository and navigate to data/
4. Run the pipeline: snakemake --cores all --configfile config.yaml --config gcp_rootdir="<bucket_name>/some_directory/"

Alternatively, in Step 4 you can submit the pipeline as a job. Create job.py containing the following:

import snakemake
snakemake.main(
	[
		"--snakefile",
		"/path/to/Snakefile",
		"--cores",
		"all",
		"--configfile",
		"/path/to/config.yaml",
		"--config",
		'gcp_rootdir="<bucket_name>/some_directory/"',
	]
)

Submit the script with hailctl dataproc submit <cluster_name> job.py

The "Analysis" pipeline

analysis/ contains scripts that calculate and visualise CAPS scores using files created in data/.

How to run

1. Navigate to CAPS/analysis/
2. snakemake --cores all --config gcp="False" (faster: uses copies from files/) or snakemake --cores all --config gcp="True" gcp_rootdir="<bucket_name>/some_directory/" (slower: uses GS files)

Using CAPS

Custom sets of variants

To get CAPS estimates for your set of variants, use the template file: snakemake -s template -c1 -C [KEY=VALUE ...]. The required values are

• obs (grouped variants annotated with at least context, ref, alt, methylation_level, singleton_count and variant_count fields)
• exp (expected proportions, one per context-ref-alt-methylation_level group)
• var (variable of interest, must be a valid field in obs)
• calculate_caps_script (calculate_caps.R)
• viz_scores_script (viz_scores.R)
• scores (filename for the output scores)
• plot (filename for the output plot)

For example, snakemake -s template -c1 -C obs=analysis/canonical_splice_site_vars.tsv exp=model/phat.tsv var=worst_csq calculate_caps_script=analysis/calculate_caps.R viz_scores_script=analysis/viz_scores.R scores=scores.tsv plot=plot.pdf.