What is the structure of the expomicset object and why use MultiAssayExperiment?
tidyexposomics wraps exposures and omics layers into a
MultiAssayExperiment, allowing for synchronized sample metadata,
flexible omics input, and pipeline tracking.
Exposure metadata lives in
colData().Omics assays live in
experiments().Row-level feature annotations live in
rowData().Results data live in
metadata().
How are DEGs defined and filtered before correlation?
You control:
logfc_thresh(log fold change)pval_thresh(adjusted p-value)score_col(stability score)score_thresh(stability filter if using sensitivity)
If you skip sensitivity, all DEGs can be used. But using stability scores makes your associations more robust by selecting features that are consistently significant across different pre-processing conditions.
How is run_sensitivity_analysis() different from just re-running DA?
sensitivity_analysis() performs systematic re-analysis
under different pre-processing assumptions, capturing:
Different scaling methods
Count/proportion thresholds
Covariate inclusion/exclusion
Different sampling using bootstrapping
It calculates a stability score per feature to measure how robust significance is across conditions — not just significance in one model or set of preprocessing conditions.
What do exposome scores represent, and when should I use them?
Exposome scores summarize multi-exposure burden into one variable using strategies like:
median: Calculates the median of the exposure variables.mean: Calculates the mean of the exposure variables.sum: Calculates the sum of the exposure variables.pca: Calculates the first principal component of the exposure variables.irt: Uses Item Response Theory to calculate the exposome score.quantile: Calculates the quantile of the exposure variables.
Use when:
You want dimensionality reduction
Or you believe co-exposure effects are more meaningful than single exposures
You can then associate them with outcomes using
run_association().
How do I interpret enrichment results from
deg_exp_cor?
Each enriched term represents a biological process or pathway that is:
Affected at the omics level (differential feature)
Correlated with an environmental exposure
This supports mechanistic interpretation of how exposures may impact disease-relevant pathways.
I see significant features with low stability — what does that mean?
If features pass significance filters but have low stability scores, they:
Might be highly sensitive to pre-processing choices
Are less reliable in real-world datasets
Use plot_sensitivity_summary() to visualize the trade-off between stability and significance.