DOLPHIN

DOLPHIN assigns domain-aware scores and extrapolated allele frequencies to improve classification of human missense variants using protein domain information.

Key Features:

• PM1 Criterion Enhancement: Increases applicability of the ACMG-AMP PM1 criterion from ~10% to ~30% by identifying protein domain residues significantly impacted by variants.
• DOLPHIN Scores: Generates residue-level scores using Pfam alignments across eukaryotes to identify critical residues within protein domains.
• Enriched Variant Frequency Data: Enriches gnomAD variant frequencies at the residue level within domains and validates frequency extrapolation using ClinVar.
• New Benign Support Criterion (BP8): Introduces BP8 as a benign-supporting criterion applicable to approximately 33.2% of variants.
• Extrapolated Frequency Data: Provides extrapolated allele frequencies for ~31.8% of variants compared with 7.6% coverage in the original gnomAD dataset.

Scientific Applications:

• Variant Classification: Facilitates application of PM1 and expands the use of PM2/BS1 criteria for classifying amino acid substitutions within protein domains.
• Pathogenic Variant Identification: Supports identification of pathogenic variants within protein domains, which cover nearly 40% of proteins.

Methodology:

DOLPHIN integrates Pfam alignments and gnomAD variant frequencies, computes residue-level scores from Pfam alignments across eukaryotes, extrapolates/enriches per-residue allele frequencies, and validates outputs against ClinVar.