Skip to content

CADD (Combined Annotation Dependent Depletion)

Resource Location

/labs/SysMedBio/Q/Resources/Genomics/CADD

/Volumes/lab/SysMedBio/Resources/Genomics/CADD

Q:\SysMedBio\Resources\Genomics\CADD

CADD is a computational tool designed to estimate the relative deleteriousness (or pathogenicity) of genetic variants throughout the human genome.

Scientific Context

Unlike tools that only evaluate specific classes of variants (e.g., non-synonymous or coding variants), CADD scores almost all possible single-nucleotide variants (SNVs) as well as short insertions/deletions (indels) across both coding and non-coding regions.

How CADD Works

  • Feature Integration: CADD integrates over 60 diverse genomic annotations (such as conservation scores, epigenetic flags, sequence context, and transcript annotations) into a single metric.
  • Training Framework: The model uses machine learning trained to distinguish between proxy-neutral variants (human alleles that have arisen and fixed in the human lineage since the split with chimpanzees, assumed to have survived natural selection) and proxy-deleterious variants (simulated de novo mutations that have not been filtered by selection).
  • The C-score: CADD outputs a Phred-scaled score (C-score). A score of 10 indicates that the variant is in the top 10% of most deleterious variants across the genome; a score of 20 indicates the top 1% (most deleterious); a score of 30 represents the top 0.1%, and so on.

Main Publications

  • Title: A general framework for estimating the relative pathogenicity of human genetic variants
  • Authors: Martin Kircher, Daniela M. Witten, Preti Jain, Brian J. O'Roak, Gregory M. Cooper, & Jay Shendure
  • Journal: Nature Genetics (2014, Vol. 46, Issue 3, Pages 310–315)

  • DOI: 10.1038/ng.2892

  • Title: CADD: predicting the deleteriousness of variants throughout the human genome

  • Authors: Philipp Rentzsch, Daniela Witten, Gregory M. Cooper, Jay Shendure, & Martin Kircher
  • Journal: Nucleic Acids Research (2019, Vol. 47, Issue D1, Pages D886–D894)
  • DOI: 10.1093/nar/gky1016

Available Files

The CADD directory contains precomputed Phred-scaled deleteriousness scores (version 1.7) for all possible single-nucleotide variants:

  • whole_genome_SNVs_GRCh37.tsv.gz: Precomputed genome-wide C-scores for all possible SNVs using the GRCh37/hg19 assembly.
  • whole_genome_SNVs_GRCh37.tsv.gz.tbi: Tabix index file for the GRCh37 SNVs dataset.
  • whole_genome_SNVs_GRCh38.tsv.gz: Precomputed genome-wide C-scores for all possible SNVs using the GRCh38/hg38 assembly.
  • whole_genome_SNVs_GRCh38.tsv.gz.tbi: Tabix index file for the GRCh38 SNVs dataset.