Is EPHA3 p.Leu668Pro (c.2003T>C) pathogenic?

EPHA3 p.Leu668Pro (c.2003T>C) is classified Uncertain significance in EPHA3 (ClinVar, Single submitter, 1 submitter; last evaluated 2024-02-05). AlphaMissense: likely pathogenic (0.9986); Ultra-rare (popmax 0.0121%, sas).

How common is EPHA3 p.Leu668Pro (c.2003T>C) in the general population?

EPHA3 p.Leu668Pro (c.2003T>C) is ultra-rare (popmax 0.0121%, sas) (gnomAD).

What do computational predictors say about EPHA3 p.Leu668Pro (c.2003T>C)?

AlphaMissense predicts likely pathogenic (score 0.9986) — an in-silico prediction, not a clinical call.

EPHA3 p.Leu668Pro

c.2003T>C · rs748606864 · VCV3089576

Uncertain significance Single submitter

Classification

Uncertain significance

ClinVar

Review

Single submitter ★☆☆☆

ClinVar · 1 submitter

AlphaMissense

likely pathogenic

0.9986 · ACMG-weighted

gnomAD v4.1

Rare

ultra-rare (popmax 0.0121%, sas)

Missense predictors

3/3 damaging

AlphaMissense, REVEL, SaProt

Computational evidence

Protein change
AlphaMissense	0.9986	likely pathogenic · top 4% in EPHA3
REVEL	0.942	strongly pathogenic-leaning
SaProt	LLR -18.28	damaging
⇒ AlphaMissense, REVEL and SaProt all call this damaging.
Splicing
SpliceAI	—	not assessed · no allele here is in the release
Constraint at this position
Conservation	phyloP 7.9	GERP 2.1, phastCons 1
Regional constraint	o/e 0.98	unconstrained · MANE · 372 kb · p 0.42 · regional, not allelic
Overall
CADD	29.7	PHRED · no single calibrated threshold — expert panels disagree

ClinVar classifies this uncertain; the computational predictors unanimously lean damaging — a triage signal, not a reclassification

AlphaMissense carries the ACMG weight for missense; REVEL and SaProt are independent agreement signals.

Population frequency

gnomAD v4.1 · exomes + genomes

Grpmax group 0.0121% South Asian · 11/90,956 alleles

Filtering AF 0.00677% ClinGen BA1/BS1 input FAF95, joint · FAF99 0.00523%

Overall 0.000754% 11/1,458,980 alleles · ultra-rare

Rarity supports pathogenicity (ACMG PM2); frequency above a disease-specific BA1/BS1 threshold argues against it. A variant frequent in one ancestry group can be near-absent in others, so a single global figure can mislead.

Variant & gene context

Gene: EPHA3
Protein: p.Leu668Pro
Coding: c.2003T>C
MANE Select: NM_005233
dbSNP: rs748606864
Type: single nucleotide variant
Location: chr3:89419319 (GRCh38)
Gene constraint: LOEUF 0.508 · pLI 0.80 (gnomAD)
Structure: AlphaFold pLDDT 81.22 · 28 PDB
Residue 668: an α-helix; the protein kinase domain

Full EPHA3 gene report on Sugi Atlas → Gene function, pathways, protein structure, interactions & more

Similar variants

EPHA3 p.Ala777Gly Benign ★★ ⚠ same type
EPHA3 p.Val597Ile Likely benign ★★ ⚠ same type
EPHA3 p.Ala182Val Uncertain significance ★ ⚠ same type
EPHA3 p.Ala427Val Uncertain significance ★ ⚠ same type
EPHA3 p.Ala546Thr Uncertain significance ★ ⚠ same type
EPHA3 p.Ala550Pro Uncertain significance ★ ⚠ same type
EPHA3 p.Ala895Val Uncertain significance ★ ⚠ same type
EPHA3 p.Arg312Trp Uncertain significance ★ ⚠ same type

Built on 18 open, versioned datasets — ClinVar, gnomAD, AlphaMissense, REVEL, ClinGen & more — plus Sugi Variant's own cross-source analysis. Source data 2026-01–2026-10, ClinVar 2026-10. Full sources & versions →

Cite Sugi Variant

Gür T. Sugi Variant: A Deterministic Per-Variant Germline Reference from a Knowledge Graph for Humans and AI Agents. 2026. doi:10.5281/zenodo.21509445. This variant: https://sugi.bio/variant/epha3-p-leu668pro (accessed [today's date]).

Preprint & method: sugi.bio/variant/method Machine-readable: this page as Markdown