Skip to contents

Wrapper function to CisTarget parameters

params_cistarget.Rd

auc_threshold and max_rank are two different cutoffs and are easy to confuse. The first truncates the recovery curve used to score motif enrichment, the second sets how deep into the ranking the background curve (mean + 2 SD across all motifs) is built, and therefore where the leading edge cuts. RcisTarget uses maxRank = 5000 and nMean = 100; the defaults here follow it. auc_threshold follows pySCENIC at 5%, RcisTarget itself uses 3%.

Usage

params_cistarget(
  auc_threshold = 0.05,
  nes_threshold = 3,
  max_rank = 5000L,
  n_mean = 100L,
  rcc_method = c("approx", "icistarget"),
  high_conf_cats = c("directAnnotation", "inferredBy_Orthology"),
  low_conf_cats = c("inferredBy_MotifSimilarity",
    "inferredBy_MotifSimilarity_n_Orthology")
)

Arguments

auc_threshold

Numeric. Numeric between 0 and 1. Proportion of genes to use for AUC threshold calculation. Default is 0.05 (5% of genes). Defaults to 0.05.

nes_threshold

Numeric. Normalised Enrichment Score threshold for significant motifs. Default is 3.0. Defaults to 3.0.

max_rank

Integer. Depth of the recovery curves used to derive the background and the leading edge. Clamped to the number of genes in the ranking database. Default is 5000, the RcisTarget value. Defaults to 5000L.

n_mean

Integer. Window for the rolling mean smoothing the background recovery curve. Only read when rcc_method = "approx". Default is 100, the RcisTarget value. Defaults to 100L.

rcc_method

String. Method for recovery curve calculation. Either "approx" (approximate, faster) or "icistarget" (exact, slower). One of c("approx", "icistarget"). Defaults to "approx".

high_conf_cats

String vector. Annotation categories considered high confidence. Default includes direct annotations and orthology-based inferences. Defaults to c("directAnnotation", "inferredBy_Orthology").

low_conf_cats

String vector. Annotation categories considered lower confidence. Default includes motif similarity-based inferences. Defaults to c("inferredBy_MotifSimilarity", "inferredBy_MotifSimilarity_n_Orthology").

Value

A named list with the following elements:

  • auc_threshold - Numeric. Numeric between 0 and 1. Proportion of genes to use for AUC threshold calculation. Default is 0.05 (5% of genes). Defaults to 0.05.

  • nes_threshold - Numeric. Normalised Enrichment Score threshold for significant motifs. Default is 3.0. Defaults to 3.0.

  • max_rank - Integer. Depth of the recovery curves used to derive the background and the leading edge. Clamped to the number of genes in the ranking database. Default is 5000, the RcisTarget value. Defaults to 5000L.

  • n_mean - Integer. Window for the rolling mean smoothing the background recovery curve. Only read when rcc_method = "approx". Default is 100, the RcisTarget value. Defaults to 100L.

  • rcc_method - String. Method for recovery curve calculation. Either "approx" (approximate, faster) or "icistarget" (exact, slower). One of c("approx", "icistarget"). Defaults to "approx".

  • high_conf_cats - String vector. Annotation categories considered high confidence. Default includes direct annotations and orthology-based inferences. Defaults to c("directAnnotation", "inferredBy_Orthology").

  • low_conf_cats - String vector. Annotation categories considered lower confidence. Default includes motif similarity-based inferences. Defaults to c("inferredBy_MotifSimilarity", "inferredBy_MotifSimilarity_n_Orthology").