How kircherlab/cadd-scripts works
CADD is an offline genome-variant scoring tool for SNVs and short InDels on GRCh37 and GRCh38. The principal workflow accepts a VCF, prepares variants and annotations, derives model features, predicts deleteriousness, and emits tabular C-scores; users can request annotated output or use pre-scored files. The README documents the command-line and workflow entry points, while sampled scripts support variant normalization, annotation, feature transformation, and score post-processing. The Snakefile’s rules were not sampled, so its internal dispatch links are shown only at the documented workflow level.
Main entry points: Snakefile Workflow engine · CADD.sh CLI wrapper · extract_scored.py Pre-score extraction · VCF2vepVCF.py Variant normalization
Components 21
Workflow entry
- CLI wrapper
CADD.sh - Workflow engine
Snakefile
Variant preparation
- Variant normalization
src/scripts/VCF2vepVCF.py - VEP consequence annotation
envs/vep.yml - VEP VCF enrichment
src/scripts/annotateVEPvcf.py
Features and models
- Genomic annotations
src/scripts/lib/Annotations.py - Genome annotation tracks
data/annotations - Track definitions
src/scripts/lib/tracks.py - Feature transformation
src/scripts/trackTransformation.py - Regulatory sequence model
src/scripts/lib/tools/regulatorySequence/predictVariants.py - Splice prediction
src/scripts/lib/tools/MMSplice.py - Protein language scores
src/scripts/lib/tools/esmScore/esm_lib.py
Scoring and output
- CADD prediction models
data/models - CADD score prediction
src/scripts/predictSKmodel.py - Variant score selection
src/scripts/max_line_hierarchy.py - PHRED conversion
src/scripts/appendPHREDscore.py - Pre-scored variants
data/prescored - Pre-score extraction
src/scripts/extract_scored.py
Other components
- CADD user
- VCF input
- CADD result table
Connections 21
- CADD user → CLI wrapper: invokes
- VCF input → CLI wrapper: provides VCF
- CLI wrapper → Workflow engine: starts workflow
- Workflow engine → Variant normalization: runs preparation
- Variant normalization → VEP consequence annotation: prepares variants
- VEP consequence annotation → VEP VCF enrichment: annotates consequences
- VEP VCF enrichment → Genomic annotations: loads annotations
- Genomic annotations → Genome annotation tracks: queries tracks
- Workflow engine → Feature transformation: runs feature processing
- Track definitions → Feature transformation: defines features
- Workflow engine → Regulatory sequence model: runs optional prediction
- Workflow engine → Splice prediction: runs optional prediction
- Workflow engine → Protein language scores: runs optional scoring
- Workflow engine → CADD score prediction: runs scoring
- CADD prediction models → CADD score prediction: provides models
- Workflow engine → Variant score selection: selects variant scores
- Workflow engine → PHRED conversion: converts scores
- Workflow engine → Pre-score extraction: uses pre-scored data
- Pre-scored variants → Pre-score extraction: provides scores
- PHRED conversion → CADD result table: writes scores
- Pre-score extraction → CADD result table: writes scores