Load a Prebuilt Biomarker Reference Distribution
Source:R/prebuilt_wss.R
load_prebuilt_biomarker_reference.RdSome prebuilt reference sets ship a baseline biomarker reference
distribution used by normalize_wss_biomarkers() (method = "project")
to rank-normalize new raw biomarker values without needing a sizeable
cohort of your own. Each biomarker's reference values are drawn from the
real baseline cohort, but independently randomly shuffled so that no
two values across different biomarkers can be attributed to the same
original subject. This preserves each biomarker's marginal distribution
exactly (every value, tie, and extreme is real) while destroying the
joint/multivariate fingerprint that carries re-identification risk.
Arguments
- prebuilt
Name of a prebuilt reference set. See
list_prebuilt_wss()for available options.
Value
A data.table with columns biomarker and value (long format,
one row per reference observation; biomarkers have differing numbers of
rows due to differing missingness in the original cohort), or NULL if
the given reference set has no associated biomarker reference.
Details
Because of this, the returned reference has no valid joint structure — inter-biomarker correlations are deliberately destroyed. Only use it for independent per-biomarker operations (rank/quantile lookups), never for anything requiring multivariate relationships (e.g. correlation or joint modeling).