Conducting Exposome-Wide Association Studies
Instructor: Chirag J Patel | DBMI Faculty Page
A hands-on course on the design, execution, and interpretation of Exposome-Wide Association Studies (ExWAS) using NHANES data and the nhanespewas R package.
Updated: February 18, 2026 | GitHub | @chiragjp
Getting Started
Prerequisites
Install the required R packages before the first class:
source("base_packages.R")The nhanespewas Package (Modules 4-10)
devtools::install_github("chiragjp/nhanespewas")Download the SQLite database from Figshare and place it in your working directory. See Module 4 for detailed setup instructions.
Bundled Data (Modules 1-3)
Modules 1-3 use a bundled NHANES dataset (data/nh.Rdata) and do not require the nhanespewas package or external database.
Modules
| # | Module | Topics | Data Source |
|---|---|---|---|
| 1 | The Exposome | Exposome definition, NHANES overview, E-G-P-D framework, ExWAS concept | nh.Rdata |
| 2 | R and Tidyverse Foundations | Pipe operator, dplyr verbs, ggplot2, broom, nest_by |
nh.Rdata |
| 3 | Statistical Foundations | Survey design, svyglm, log-transform, Z-score, FDR, DAGs, per-association confounding |
nh.Rdata |
| 4 | The nhanespewas Package | Installation, database, catalogs, adjustment models, pe_flex_adjust() |
nhanespewas |
| 5 | Conducting an ExWAS | ExWAS loop, tryCatch, FDR, replication, parallelization, logistic ExWAS |
nhanespewas |
| 6 | Interpreting Results | Volcano plots, delta R-squared, replication tables, sensitivity analysis, STROBE-E | nhanespewas |
| 7 | Advanced Topics | Meta-analysis (UWLS), interaction testing, confounding mitigation (negative controls, MR, triangulation, DML/TMLE) | nhanespewas |
| 8 | Putting It All Together | The PE Atlas (Patel, Ioannidis, Manrai — Nature Medicine), 120K associations, exposome vs. genome, poly-exposure scores | nhanespewas |
| 9 | Untargeted Exposomics | LC-HRMS, blood exposome, annotation challenge, untargeted ExWAS, batch effects | Conceptual |
Assignments
Homework
| Assignment | Focus | Code Required? |
|---|---|---|
| Assignment 1: Confounding, Adjustment, and Interpretation | DAGs, volcano plot interpretation, adjustment model reasoning | No |
| Assignment 2: Validation and Multiple Testing | BH-FDR by hand, triangulation design, MR feasibility, negative controls | Minimal |
Course Project
| Course Project: Your Own ExWAS | Select or derive a phenotype, run a full ExWAS, interpret top hits with exposure-specific DAGs, design a validation strategy |
The project is weighted 70% interpretation, 30% execution. A null result is a valid outcome if interpreted thoughtfully. Use of AI tools is permitted but must be cited.
Key References
- Patel CJ, Bhattacharya J, Butte AJ. An Environment-Wide Association Study (EWAS) on Type 2 Diabetes Mellitus. PLoS ONE 2010; 5(5):e10746.
- Patel CJ, et al. A database of human exposomes and phenomes from the US National Health and Nutrition Examination Survey. Scientific Data 2016; 3:160096.
- Patel CJ, Ioannidis JPA, Manrai AK. An atlas of exposome-phenome associations in health and disease risk. Nature Medicine, in press.
- Chung MK, et al. The exposome and exposome-wide association studies. Exposome 2024.
- Munafo MR, Davey Smith G. Robust research needs many lines of evidence. Nature 2018; 553:399-401.
- Vermeulen R, Schymanski EL, Barabasi AL, Miller GW. The exposome and health: where chemistry meets biology. Science 2020; 367:392-396.
External Resources
| Resource | Link |
|---|---|
nhanespewas R package |
https://github.com/chiragjp/nhanespewas |
| PE Atlas (interactive) | https://pe.exposomeatlas.com |
| NHANES (CDC) | https://www.cdc.gov/nchs/nhanes/index.htm |
| Full cohort database | https://doi.org/10.6084/m9.figshare.29182196 |
| Summary statistics | https://doi.org/10.6084/m9.figshare.29186171 |
Supported By
This course is supported by the National Institutes of Health (NIH):
- National Institute of Environmental Health Sciences (NIEHS): R01ES032470, U24ES036819
- National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK): R01DK137993