title	My_Project.md
author	First Last
date	2023-12-01
output	github_document

Cross-ancestry impact of genotype-environment interactions and social determinants of health on AD endophenotypes

Background/Overview

Variations in phenotypic outcomes are influenced by the dynamic interplay between genotype and environment interactions (GxE). GxE recognizes that the traits and diseases are not solely dictated by genetic makeup or environmental exposures but rather by the interactions between the two. These exposures can include behavioral and lifestyle choices such as exposure to toxins/pollutants, diet, physical activity, and clinical factors like comorbid conditions, all of which differs across populations and environments. Additionally, SDoH play an external role in GxE by influencing how environmental factors modulate genetic predispositions to diseases. PRS hold a great promise for precision medicine; however, GxE reduces the accuracy of PRS even within ancestral groups, along with differences in linkage disequilibrium and allele frequency.

Objectives

Aim 1: Evaluate the predictive accuracy of AD-PRS computed using single- and cross-ancestry PRS methods in diverse populations -Evaluate associations of clinical risk score with AD endophenotypes

Aim 2: Evaluate the associations of genetic, environmental, and social factors with AD endophenotypes

Aim 3: Integrate genetic and clinical risk factors to assess genomic-informed dementia risk

Analysis Plan

Genetic Ancestry: Individual ancestry will be assigned by calculating principal components using a reference panel of the 1000 Genome Project (1KG). The HABS-HD samples will then be projected onto the resulting PC space. The ancestry for each individual will be determined by the nearest reference population based on Euclidian distance to the reference population centers.

PRS: AD-PRS will be constructed using PRS-CSx using ancestry-specific GWAS summary statistics of European (Bellenquez et al. 2022), African American (Kunkle et al. 2021), East Asian (Shegemizu et al. 2021), and Caribbean Hispanic (Ghani et al. 2012) cohorts. The resulting scoring file will then be used to generate ancestry-normalized PRS for HABS-HD samples using the polygenic score catalog calculator (PSCG). PGSC is a reproducible pipeline that calculates PRS by utilizing a scoring file and normalizing for genetic ancestry.

Clinical Risk Scores: Clinical risk scores (CRS) for AD will be calculated for each HABS-HD participant following the Washington Heights-Inwood Columbia Aging Project (WHICAP) scoring. The score is the sum of the points from risk factors including: age, sex, ethnicity, presence of diabetes and/or hypertension, current smoking status, HDL level, BMI, education, and APOE allele status [8]. The scores will be standardized for the regression model.

Regression model: Variables will be incorporated into regression models in a stepwise manner to evaluate their associations with AD endophenotypes, including cognitive function, ß-amyloid level, hyperphosphorylated tau level, and neuroimaging. Associations of covariates will be assessed using odds ratios, and the model performance will be assessed using Nagelkerke R2 and AUC.

Data Preprocessing

Refer to standardization.qmd

Data Structure

project_directory # The working directory
├── data
│   ├── raw_data.csv 
│   └── processed_data.csv 
├── output
│   ├── results 
│   │     ├── results_aim_1.csv # write a note about what this file contains.
│   │     └── results_aim_2.csv
│   └── visualization 
│         └── plots_aim_1.png
│         └── plots_aim_2.png
└── scripts 
    ├── data_processing.Rmd
    ├── analyses_aim_1.1.sh
    ├── analyses_aim_1.2.Rmd
    ├── analyses_aim_2.1.py
    └── data_visualziation.Rmd

Expected Results

We anticipate higher CRS, unfavorable PRS, and disadvantaged SDoH to be associated with higher risk of AD. We also anticipate disadvantaged SDoH and higher CRS to exacerbate risk from AD-PRS. We will further investigate whether the associations of genetic, environmental, and social determinants with AD differ across populations and how GxE may change in different populations. This study aims to answer the pressing questions of the effects of various risk factors on AD risk across diverse populations and to identify risk groups and modifiable risk factors for early therapeutic interventions. This research will contribute to the development of precision medicine approaches that tailor prevention and treatment strategies to individual risk profiles, ultimately improving health outcomes for diverse populations.

Responsibilities and Acknowledgements