Doublet detection with boosted doublet classification — doublet_detection_boost

This function implements the boosted doublet detection. It generates through several iterations simulated doublets, generate kNN graphs, runs Louvain clustering and assesses how often an observed cells clsuters together with the simulated doublets.

Usage

doublet_detection_boost_sc(
  object,
  boost_params = params_boost(),
  seed = 42L,
  streaming = FALSE,
  .verbose = TRUE
)

Arguments

object: SingleCells class.
boost_params: A list with the final scrublet parameters, see params_boost() for full details.
seed: Integer. Random seed.
streaming: Boolean. Shall streaming be used during the HVG calculations. Slower, but less memory usage.
.verbose: Boolean. Controls verbosity of the function.

Value

A boost_res class that has with the following items:

predicted_doublets - Boolean vector indicating which observed cells predicted as doublets (TRUE = doublet, FALSE = singlet).
doublet_scores_obs - Numerical vector with the likelihood of being a doublet for the observed cells.
voting_avg - Numerical vector with the average voting score.