Inspect detection-limit cap activations from impute_chemistry()

An imputed below-detection (BDL) cell must not exceed its detection limit (DL). The posterior prediction is not itself constrained below the DL, so impute_chemistry() caps any imputed BDL cell whose estimate came out above the limit (bdl_cap = TRUE). Frequent capping signals tension between the modelled chemistry and the reported limits, so the cells that triggered the cap are worth auditing rather than trusting blindly.

Usage

bdl_cap_summary(x)

Arguments

x

A data frame returned by impute_chemistry().

Value

A tibble with one row per (sample_id, analyte) cell that exceeded its detection limit, with columns detection_limit, n_rows (rows over the DL — one per draw when return = "draws"), max_imputed, max_ratio (max_imputed / detection_limit) and capped (whether the cap was applied). The tibble carries the overall fire_rate (fraction of capable BDL cells that were clipped) and n_bdl_cells as attributes, also reported via a message. Returns NULL invisibly when no cell exceeded its DL.

Details

impute_chemistry() attaches a per-cell audit summary to its result as the "bdl_cap_summary" attribute; this accessor returns it. Because plain attributes are dropped by most dplyr verbs, call this on the frame as returned by impute_chemistry(), before further wrangling.

See also

impute_chemistry()