Research questions:
What germline genetic variants drive tissue-specific aging phenotypes, and can polygenic scores derived from histopathological analyses predict age-related disease risk at population scale?
Background:
We have developed deep learning methods to quantify tissue-specific biological age and microanatomical organization from histopathological images across 40 human tissues (GTEx cohort, n=940). These “histophenotypes” capture age-associated cellular and structural changes with unprecedented granularity. However, cohorts of histopathological data often lack the statistical power for robust genetic discovery and longitudinal health outcomes that the UK biobank provides.
Objectives:
1. Perform GWAS on histopathological phenotypes (biological age acceleration, microanatomical domain abundance, cellular morphology) in GTEx to identify tissue-specific genetic associations.
2. Construct polygenic scores (PGS) from these associations and transfer them to UK Biobank participants, effectively inferring tissue-specific aging phenotypes via germline genetics – even without histological data.
3. Validate PGS by testing associations with UK Biobank clinical outcomes, demographic factors, behavioral variables, and plasma proteomics-derived biological age measures.
4. Evaluate whether tissue-specific genetic aging signatures predict disease incidence and mortality.
Rationale:
Current aging research lacks mechanistic understanding of how genetics influences tissue-specific structural decline. By integrating spatially-resolved histopathological phenotypes with large-scale genetic data, we create a framework for localizing genetic effects within tissues. The UK Biobank’s scale, comprehensive phenotyping, and longitudinal follow-up enable validation and clinical translation impossible in smaller cohorts. This approach moves from descriptive aging maps to causal, genetically-informed models connecting germline variants, tissue architecture, and pathology.