bushlab/cadd-scoring repository overview

This is a container for the purpose of running Combined Annotation Dependent Depletion (CADD) v1.6, which is a tool for scoring the deleteriousness of human genetic variants.

Please refer to official documentation⁠ before using this container.

⁠Downloading necessary files

CADD requires annotated human genome builds (GRCh37 and GRCh38) for offline running of scripts. CADD also has prescored files for GRCH37 and 38.

These files can be downloaded here.⁠

Please note that these files are very large.

⁠Mounting downloaded files

⁠Mounting the annotated human genome builds

The annotated human genome builds need to be downloaded, unzipped, and mounted to:

/CADD-scripts/data/annotations

in the container.

⁠Mounting annotated prescored files

Prescored files need to be mounted with their respective tabix indices.

Annotated prescored files need to be mounted to:

/CADD-scripts/data/prescored/${GENOME_BUILD}_${VERSION}/anno

For example,

/CADD-scripts/data/prescored/GRCh38_v1.6/anno

⁠Mounting non-annotated prescored files

Non-annotated prescored files need to be mounted to:

/CADD-scripts/data/prescored/${GENOME_BUILD}_${VERSION}/no_anno

For example,

/CADD-scripts/data/prescored/GRCh38_v1.6/no_anno

⁠Docker Run Command

The easiest way to run CADD is to download the annotated human genome builds, and then mount them to /CADD-scripts/data/annotations.

Once you have the annotated human genome builds downloaded, the container can be run in interactive mode with the following command:

docker run -it -v /path/to/annotated/builds:/CADD-scripts/data/annotations bushlab/cadd-scoring

Where path/to/annotated/builds is the path to the folder that contains the unzipped annotated human genome builds.

⁠References

CADD is Copyright (c) University of Washington, Hudson-Alpha Institute for Biotechnology and Berlin Institute of Health 2013-2020. All rights reserved. The purpose of this docker image is to provide a resource for easy use of CADD.

Rentzsch P, Schubach M, Shendure J, Kircher M.
CADD-Splice—improving genome-wide variant effect prediction using deep learning-derived splice scores.
Genome Med. 2021 Feb 22. doi: 10.1186/s13073-021-00835-9.
PubMed PMID: 33618777.