The Human Phenome Research Group focuses on the systematic characterization of human phenotypic diversity. Our research covers multiple dimensions of human phenotypes, including skin appearance, limb and organ imaging, aging processes, and environmental exposures. We center our work on phenotype quantification, biological mechanism elucidation, and prediction of disease and mortality risk.

Our overarching research goals are to:

(1) develop scalable and reproducible human phenotype quantification frameworks by leveraging artificial intelligence and advanced statistical methodologies;

(2) systematically dissect the genetic and environmental determinants of human phenotypic variation through the integration of multi-layer data, including genetics, epigenetics, proteomics, metabolomics, imaging omics, and exposomics; and

(3) characterize the heterogeneity of complex traits (including disease susceptibility) at the population level, uncover their biological foundations, and develop precise predictive models, thereby establishing methodological and theoretical foundations for personalized health management.

The major research directions of the group include:

1. Quantification of Skin and Skin Appendage Phenotypes and Determinant Analysis

The skin is the largest organ of the human body and serves as a primary interface with the external environment. However, prior studies have predominantly focused on skin diseases, while systematic investigations of normal skin phenotypes and their population-level variation remain limited.

Our research in this area includes:

I.        Development of quantitative methods for skin and skin appendage phenotypes.

We develop high-throughput, image-based phenotype quantification methods to precisely characterize skin features such as wrinkles, pigmentation spots, hair density, and sweat gland distribution. By integrating computer vision and deep learning techniques, we aim to improve measurement accuracy, robustness, and cross-population comparability.

II.      Genetic and environmental determinants of skin phenotypes.

We apply genome-wide association studies (GWAS) to identify genetic variants associated with skin phenotypes, and integrate environmental exposure factors—including ultraviolet radiation, air pollution, and tobacco smoke—to systematically investigate gene–environment interactions underlying skin phenotype variation.

III.    Molecular mechanisms underlying skin phenotype variation.

By integrating multi-omics data—including DNA methylation, transcriptomics, proteomics, metabolomics, skin microbiome profiles, and single-cell sequencing—we explore the molecular basis of inter-individual skin phenotype differences, providing mechanistic insights into skin health maintenance and disease risk.

2. Biological Understanding of Imaging-Based Phenotypes and Determinant

Traditional phenotype measurements rely heavily on manual assessment, which is subjective, poorly reproducible, and difficult to scale in large population studies. Imaging data offer high-dimensional and high-fidelity representations of human phenotypes, and when combined with artificial intelligence, enable precise and scalable phenotype quantification.

Our research directions include:

I.        Quantification of imaging phenotypes.

Using deep learning techniques, we extract craniofacial, limb and brain imaging-derived phenotypes (IDPs) from 2D/3D images, CT, MRI and DXA.

II.      Genetic and non-genetic factors underlying individual phenotypic variation.

By integrating multi-omics data, we investigate genetic and environmental factors affecting IDPs.

III.    Developmental biology interpretation.

We understand variation in IDPs from the perspective of developmental biology by leveraging single-cell RNA-sequencing and spatial transcriptomics data. We study the evolutionary mechanisms of IDPs driven by natural selection, ancient DNA introgression and primate evolution using evolutionary analyses.

3. Interpretable Aging Quantification and Its Determinants

Aging is a complex, multi-scale process spanning molecular, cellular, tissue, and organ levels, ultimately manifesting as increased disease risk and mortality. Developing scientifically grounded and interpretable aging metrics is a central challenge for understanding functional decline, predicting health risk, and evaluating intervention effects.

Our research includes:

I.        Theoretical modeling and development of aging quantification methods.

Although numerous aging metrics have been proposed, their theoretical foundations and biological interpretations remain an active area of investigation. We aim to construct a more coherent and interpretable theoretical framework for aging quantification by integrating definitions of healthy aging from the World Health Organization with mathematical modeling and evolutionary theory, and to guide the development of aging metrics accordingly.

II.      Determinants of aging and population heterogeneity.

Aging is shaped by both intrinsic biological processes and extrinsic environmental and lifestyle factors. By integrating genetic information, multi-omics data, lifestyle variables, and environmental exposures, we seek to identify determinants of aging rate and patterns, characterize distinct aging subtypes, and provide a basis for precision anti-aging interventions.

III.    Aging, disease and mortality risk, and evaluation of interventions.

A key advantage of aging quantification lies in its ability to assess health status and future risk before disease onset or death. We systematically investigate associations between aging metrics and disease incidence and mortality risk, reconstruct individual aging trajectories, develop predictive models for health and mortality outcomes, and evaluate the effectiveness of potential intervention strategies.

4. Exposomics and Multi-Omics Analysis of Human Health

Exposomics focuses on environmental exposures experienced across the human lifespan, including chemical, physical, biological, and social factors, and their long-term impacts on health. Advances in high-throughput omics technologies and artificial intelligence have created new opportunities to systematically characterize exposure-health relationships.

Our research directions include:

I.        Identification of health-promoting and disease-driving exposures.

Using causal inference and statistical modeling approaches, we distinguish protective exposures from harmful, disease-driving factors, and explore personalized exposure combinations that may optimize health outcomes (e.g., diet, air quality, and social environment).

II.      Multi-omics mechanisms mediating individual health differences induced by environmental exposures.

By integrating genomic, epigenomic, proteomic, metabolomic, and microbiome data, we elucidate how environmental exposures induce heterogeneous molecular responses across individuals and how these multi-omics alterations mediate differential disease risks. By integrating exposure data with molecular and clinical phenotypes, we uncover the biological pathways and regulatory mechanisms underlying individual health differences.

III.    Development of individualized susceptibility indices for environmental exposures based on multi-omics signatures.

This direction aims to integrate multi-omics molecular signatures associated with environmental exposures to construct quantitative indices of individual susceptibility. These indices are designed to capture inter-individual heterogeneity in exposure-related disease risk and to enable risk stratification and validation across independent population cohorts.