Run Correlation Analysis — run_correlation • tidyexposomics

Computes correlations between exposures and feature types including DEGs, omics, latent factors, top factor features, or principal components (PCs). Optionally computes feature-feature correlations to support network analysis.

Usage

run_correlation(
  exposomicset,
  feature_type = c("degs", "omics", "factors", "factor_features", "exposures", "pcs"),
  exposure_cols = NULL,
  variable_map = NULL,
  n_pcs = NULL,
  feature_cors = FALSE,
  robust = FALSE,
  score_col = "stability_score",
  score_thresh = NULL,
  correlation_method = "spearman",
  correlation_cutoff = 0.3,
  cor_pval_column = "p.value",
  pval_cutoff = 0.05,
  deg_pval_col = "adj.P.Val",
  deg_logfc_col = "logFC",
  deg_pval_thresh = 0.05,
  deg_logfc_thresh = log2(1.5),
  batch_size = 1500,
  action = c("add", "get")
)

Arguments

exposomicset: A MultiAssayExperiment object.
feature_type: Type of features to correlate. One of "degs", "omics", "factors", "factor_features", "exposures", or "pcs".
exposure_cols: Optional vector of exposure column names (from colData) to use.
variable_map: Optional mapping of features to include by assay for omics mode.
n_pcs: Number of PCs to use when feature_type = "pcs".
feature_cors: Logical; if TRUE, compute correlations between features rather than with exposures.
robust: Logical; restrict DEGs to those passing sensitivity threshold.
score_col: Column name in sensitivity analysis with feature stability score.
score_thresh: Threshold for filtering robust features.
correlation_method: One of "pearson", "spearman", or "kendall".
correlation_cutoff: Minimum absolute correlation to retain.
cor_pval_column: Column in output to filter by p-value (default: "p.value").
pval_cutoff: Maximum p-value or FDR threshold to retain a correlation.
deg_pval_col: Column with DEG adjusted p-values.
deg_logfc_col: Column with DEG log fold-changes.
deg_pval_thresh: P-value cutoff for DEGs.
deg_logfc_thresh: Log fold-change cutoff for DEGs.
batch_size: Number of features to process per batch (default: 1500).
action: Whether to "add" results to metadata or "get" as a data frame.

Value

If action = "add", returns updated MultiAssayExperiment with results added to metadata. If action = "get", returns a tidy data.frame of correlations.

Examples

# create example data
mae <- make_example_data(
    n_samples = 10,
    return_mae = TRUE
)
#> Ensuring all omics datasets are matrices with column names.
#> Creating SummarizedExperiment objects.
#> Creating MultiAssayExperiment object.
#> MultiAssayExperiment created successfully.

# run correlation analysis
mae <- mae |>
    run_correlation(
        feature_type = "exposures",
        exposure_cols = c("exposure_pm25", "exposure_no2", "age", "bmi")
    )