Compute LMF for a single sample — compute_lmf_for

Internal function called once per sample by add_lmf. Computes the inverse-variance weighted mixing fraction and associated quality metrics, applying admission and quality gates before returning.

Usage

compute_lmf_for_sample(
  sample_wide,
  endmembers,
  high_info_ions,
  min_high_info_ions,
  rsd_default,
  max_sigma,
  max_chi2_df,
  uuid_lmf,
  robust_iterations = 3L,
  robust_threshold_k = 2.5,
  verbose = FALSE,
  datetime_sample = NA_character_,
  censored_wide = NULL
)

Arguments

sample_wide: One-row wide-format tibble with one column per measured (collapsed) panel ion.
endmembers: Tibble with columns ion, R, sigma_R, n_ref, L. One row per ion with usable end-member data for this feature.
high_info_ions: Character vector of ion names classified as high-information at calibration time.
min_high_info_ions: Minimum count of high_info_ions that must be present and non-NA in this sample.
rsd_default: Default relative analytical uncertainty (fraction).
max_sigma: Maximum permitted sigma_lmf.
max_chi2_df: Maximum permitted chi-squared per degree of freedom.
uuid_lmf: UUID of the LMF analyte entry in analyteDF.
robust_iterations: Number of Huber reweighting passes. Inherited from add_lmf.
robust_threshold_k: Huber threshold in standard-deviation units on studentized residuals. Inherited from add_lmf.
verbose: If TRUE, emits a per-ion diagnostic table via cli::cli_inform() showing observed concentration, end-member values, per-ion mixing fraction, original and robust weights. Inherited from add_lmf. Default FALSE.
datetime_sample: The sample datetime as a character string, used to label the per-ion diagnostic table when verbose = TRUE. Default NA_character_.
censored_wide: One-row wide-format tibble of logicals (from collapse_censored), one column per collapsed panel ion, TRUE where that ion's value was substituted at half the detection limit. Missing/unmatched ions are treated as not censored.

Value

A one-row tibble. On success: value is the robust LMF estimate, quantified = TRUE. On failure: value = NA, quantified = FALSE, lmf_reason carries a descriptive reason code. All paths return identical column structure.