Human pigmentation shows remarkable diversity, yet the genetic and evolutionary foundations of retinal pigmentation have remained poorly understood. In this study which published on Science Advances, researchers combined deep learning with large-scale genomics to systematically dissect the genetic architecture of retinal pigmentation.
They developed DeepGRP, a deep learning framework that quantifies retinal pigmentation from high-resolution color fundus images, overcoming limitations of traditional RGB-based measures that are sensitive to imaging conditions. Applying DeepGRP to genome-wide association analyses revealed 42 genetic signals associated with retinal pigmentation, including 26 previously unidentified loci, and estimated a SNP-based heritability of 21.4%, underscoring the trait’s strong genetic basis.
To link genetic associations with biological mechanisms, the authors integrated single-nucleus chromatin accessibility and RNA sequencing data from human fetal retinal tissues. These analyses highlighted key cellular contributors – particularly retinal pigment epithelium (RPE) and photoreceptor cells – and identified ARHGAP18 as a novel regulator of melanogenesis in the retina. Beyond molecular insights, evolutionary analyses provided evidence of polygenic adaptation in European populations, likely driven by selective pressures from snow-reflected light at high latitudes.
Clinically, a polygenic risk score derived from retinal pigmentation was associated with a 4.8-fold increased risk of myopia and a 1.5-fold reduced risk of skin cancer, linking retinal pigmentation to broader health outcomes. Overall, this work demonstrates the power of deep learning–based ocular phenotyping in population-scale studies and offers new insights into how genetics, cell biology, and evolution shape human retinal pigmentation and disease risk.
Research article: Genome-wide association study reveals genetic architecture and evolution of human retinal pigmentation
Unlocking Human Insight with Organoid Models
Lambda Biologics provides advanced organoid technology and services for biomedical research and drug development. We specialize in physiologically relevant 3D organoid models that bridge the gap between traditional cell cultures and animal models.
Our comprehensive solutions span disease modeling, drug screening, personalized medicine, and regulatory-compliant testing – supporting pharmaceutical companies, biotech firms, academic institutions, and cosmetics manufacturers worldwide.
Cerebral Organoid | Midbrain Organoid | Skin & Hair Organoid | Intestine Organoid | Gastric Cancer Organoid | Breast Cancer Organoid | Hepatocarcinoma Cancer Organoid | Pancreatic Cancer Organoid
