The genetics of face shape
The largest facial GWAS, the loci with an assigned facial trait, and why face shape is a continuum to blend rather than a set of gene switches. Every figure attributed.
Horizontal bars grow from a zero line, one per row. PAX3, nasion position: 1.33%; EDAR, chin protrusion and mandible length: 1.32%; RUNX2 / SUPT3H, nose bridge breadth: 1.18%; GLI3, nose wing breadth: 1.15%; DCHS2, nose protrusion and tip angle: 0.63-1.08%; Whole face, all variants (face-PVE): 4.5%.
Genetics
Face shape is highly heritable but extremely polygenic, and even the largest studies explain only a single-digit percentage of the variation. The practical consequence: facial variation is a broad continuum to be blended, not a set of gene switches to toggle.
| Result | Value |
|---|---|
| Discovery sample | 11,662 Europeans · 946 traits from 44 landmarks |
| Significant variants | 253 SNPs across 188 loci (64 SNPs at 62 novel loci) |
| Variance explained, per trait | 2.25% - 7.9% |
| Variance explained, whole face | 4.5% (face-PVE) |
| Highest / lowest region | upper nose 6.0% · lower nose 5.6% · lower cheek 3.2% |
| SNP-based heritability | avg 0.23, range 0.06-0.36 (higher: nose & forehead; lower: mouth & cheek) |
| Cross-ancestry replication | 70% of SNPs in 9,674 Chinese |
| Nose re-identification model | 382-SNP, AUC 0.67 |
Largest facial GWAS to date (Nature Communications, 2025)[2]
Gene → trait map
| Gene / locus | Trait it moves | Variance |
|---|---|---|
| PAX3 · 2q35 | nasion position (bridge root) | 1.33% |
| EDAR · 2q12 | chin protrusion, mandible length | 1.32% |
| DCHS2 · 4q31 | columella inclination, nose protrusion, tip angle | 0.63-1.08% |
| RUNX2 / SUPT3H · 6p21 | nose bridge breadth | 1.18% |
| GLI3 · 7p13 | nose wing breadth | 1.15% |
| PAX1 · 20p11 | nose-wing region | - |
Individual loci with an assigned facial trait[5]
EDAR V370A is the standout: the derived allele - common in East Asians and Native Americans, absent in Europeans and Africans - reduces chin protrusion and shortens the mandible, while also changing hair thickness, incisor shovelling and sweat-gland density. One variant, many features.[5]
Named loci worth knowing
- POU3F3 (2q12.1) - strong nose-region signal in the 2025 study.[2]
- PAX3 - nasion position / interorbital septum.[3]
- PRDM16 - nose width & height; SOX9 - nose shape; EDAR - East-Asian hair/nose traits.[3][4]
- TP63, C5orf50, COL17A1 - midface and eye-spacing measures.[3]
- HOXD cluster, BMP4 - overall proportions and jaw width.
Sources
- Shaffer et al. (2016). A GWAS Identifies Five Loci Influencing Facial Morphology. (male noses ~4 mm larger; high heritability, many small-effect variants). pmc.ncbi.nlm.nih.gov/articles/PMC3441666/
- Combined GWAS of facial traits in Europeans (2025). Nature Communications. link.springer.com/article/10.1038/s41467-025-61761-7
- Shaffer et al. / PAX3, PRDM16, TP63, C5orf50, COL17A1 (PLoS Genetics 2016). journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1006149
- Identification of five novel genetic loci related to facial morphology (SOX9, WDR35, HOXD1-MTX2). BMC Genomics 2018. link.springer.com/article/10.1186/s12864-018-4865-9
- A genome-wide association scan implicates DCHS2, RUNX2, GLI3, PAX1 and EDAR in human facial variation. Nature Communications 2016. www.nature.com/articles/ncomms11616