tailor_predict: tailor_predict

In matei-ionita/Tailor: Gaussian mixture modeling for heavy-tailed distributions in Flow Cytometry data

Description

Takes as input a tailor object and some data (could be the data used to learn the tailor object, or some new data). Computes, for each event, the mixture component from which it is most likely drawn, then maps this mixture component to its corresponding categorical cluster.

Usage

tailor_predict(
  data,
  tailor_obj,
  n_batch = 64,
  parallel = FALSE,
  verbose = FALSE
)

Arguments

data	A flowSet, flowFrame or a matrix containing events along the rows, markers along columns.
tailor_obj	A tailor object containing information about mixture components and categorical clusters. Can be obtained as the output of tailor.learn.
n_batch	A naive implementation would need nrow(data)*mixture_components memory. To reduce memory usage, process data in batches.
parallel	Boolean flag; if true, uses multithreading to process batches in parallel. For optimal runtime, if parallel = TRUE, n_batch should be a multiple of the number of cores available, as returned by parallel::detectCores().
verbose	Boolean flag; if true, outputs timing and milestone information.

Value

Two atomic vectors of integers, one giving the mixture component, and the other the categorical cluster, for each event.

Examples

fileName <- system.file("extdata", "sampled_flowset_old.rda",
                        package = "Tailor")
load(fileName)
tailor_params <- flowCore::colnames(fs_old)[c(7:9, 11:22)]
tailor_obj <- tailor_learn(data = fs_old,
                          params = tailor_params,
                          mixture_components = 50)
tailor_pred <- tailor_predict(fs_old, tailor_obj)

matei-ionita/Tailor documentation built on Jan. 4, 2021, 11:47 a.m.