DANN

Uses deep neural networks to score the deleteriousness of genetic variants.

Key Info

• URL: https://github.com/michoel-lab/dann BROKEN — repo does not exist
• Alt URL: https://cbcl.ics.uci.edu/public_data/DANN/ (pre-computed scores)
• Category: VEP
• API: Pre-computed scores (not standalone repo)
• Free: Free
• ACMG Relevance: PP3
• Link Status: BROKEN 2026-04-09 — DANN scores pre-computed, distributed via UCI and dbNSFP. Not a standalone GitHub repo.

Why It Matters

Provides a non-linear counterpart to CADD, often yielding better separation between pathogenic and benign alleles.

How to Use

DANN scores available as pre-computed whole-genome annotations. Access via dbNSFP or download from https://cbcl.ics.uci.edu/public_data/DANN/. Already tested on STRC E1659A: score 0.9946 (highly deleterious).

STRC Research Usage

Documented. Requires local installation or specialized access for STRC E1659A testing.

Next Steps

• Test with STRC E1659A variant
• Verify API access
• Document specific results

Connections

• STRC Variant c.4976A>C — Misha [see-also]
• STRC Gene [about]