Skip to contents

Applies a simple rule-based classifier to each protein using per-contrast statistics extracted by interpret_protein_factorial. Intended for BioID-style designs with average treatment, interaction, and baseline (between-cell) contrasts.

Usage

classify_factorial_batch(
  DA_results,
  id_vector = NULL,
  contrast_map = c(Avg_Treat_Bio_vs_DMSO = "Avg_Treat_Bio_vs_DMSO",
    Interaction_CHLA_vs_SKNF1 = "Interaction_CHLA_vs_SKNF1", CHLA90_Bio_vs_SKNF1_Bio =
    "CHLA90_Bio_vs_SKNF1_Bio", CHLA90_DMSO_vs_SKNF1_DMSO = "CHLA90_DMSO_vs_SKNF1_DMSO"),
  p_thresh = 0.05,
  lfc_thresh = 1
)

Arguments

DA_results

A DAList-like object with $results and optionally $annotation.

id_vector

Optional character vector of protein IDs to classify. If NULL, the function tries unique(DA_results$annotation$uniprot_id).

contrast_map

Named character vector mapping display labels to keys in DA_results$results. Defaults match the typical factorial setup.

p_thresh

Numeric. FDR threshold for significance tests. Default 0.05.

lfc_thresh

Numeric. Absolute log2 fold-change threshold used to decide whether the baseline contrast is near zero. Default 1.

Value

A data.frame with columns:

  • id - protein ID,

  • class - assigned class label,

  • base_logFC, base_q - baseline contrast statistics,

  • treat_logFC, treat_q - average treatment statistics,

  • int_logFC, int_q - interaction statistics.

Details

Classes returned include (non-exhaustive):

  • "Shared_Tx" - significant average treatment effect; non-significant interaction.

  • "WTPreferential_Tx" - significant average treatment; interaction significantly negative; baseline near zero.

  • "WTPreferential_Tx_withBaselineShift" - as above but with significant baseline shift.

  • "TruncationGain" - significant average treatment; interaction significantly positive; baseline near zero.

  • "TruncationGain_withBaselineShift" - as above but with significant baseline shift.

  • "BaselineDifferenceOnly" - significant baseline difference; average treatment not significant.

  • "Unclassified" - none of the above rules matched.

Examples

if (FALSE) { # \dontrun{
cls <- classify_factorial_batch(DA_results)
table(cls$class)

# Use stricter thresholds
cls2 <- classify_factorial_batch(DA_results, p_thresh = 0.01, lfc_thresh = 1.5)
} # }