determineDropoutCandidates: Determine Dropout Candidates
In VCCRI/CIDR: Clustering through Imputation and Dimensionality Reduction

Description Usage Arguments Details Value Examples

determines which entries in a single-cell RNA-Seq dataset are dropout candidates.

## S4 method for signature 'scData'
determineDropoutCandidates(object, min1 = 3, min2 = 8,
  N = 2000, alpha = 0.1, fast = TRUE, zerosOnly = FALSE,
  bw_adjust = 1)

`object`	the scData class object.
`min1, min2`	technical parameters used in estimating the minimum point between the first two modes of the density curve of logTPM for each cell.
`N`	number of cells to consider when determining the threshold value for dropout candidates; used in conjunction with the `fast` parameter.
`alpha`	a cutoff quantile in the range [0,1]. Thresholds outside this will be adjusted to the quantile boundary.
`fast`	Boolean; if `TRUE` (default - unless `tagType` is `cpm`), implements a fast version for datasets with a sample size greater than N. NOTE: set to `FALSE` if `tagType` is `cpm`.
`zerosOnly`	Boolean; if `TRUE`, only zeros are considered as dropout candidates; by default `FALSE`.
`bw_adjust`	bandwidth adjustment factor; CIDR uses the default bandwidth selection method ‘nrd0’ in the kernel density estimation; see `stats::density` help page for more details.

populates a Boolean matrix with the same dimension as nData. The value is TRUE for an entry if it is a dropout candidate; otherwise the value is FALSE.

an updated scData class object with the following attributes updated

`dThreshold`	a vector of library dependent dropout candidate thresholds.
`dropoutCandidates`	a matrix with the same dimension as nData. The value is `TRUE` for an entry if it is a dropout candidate; otherwise the value is `FALSE`.