Select_Model_short_TSCAN: Using clustering results (from TSCAN) based on feature...

In suke18/FEAST: FEAture SelcTion (FEAST) for Single-cell clustering

Description

Using clustering results (from TSCAN) based on feature selection to perform model selection.

Usage

Select_Model_short_TSCAN(
  Y,
  cluster,
  minexpr_percent = 0.5,
  cvcutoff = 1,
  tops = c(500, 1000, 2000)
)

Arguments

Y	A gene expression matrix
cluster	The initial cluster labels NA values are allowed. This can directly from the Consensus function.
minexpr_percent	The threshold used for processing data in TSCAN. Using it by default.
cvcutoff	The threshold used for processing data in TSCAN. Using it by default.
tops	A numeric vector containing a list of numbers corresponding to top genes; e.g., tops = c(500, 1000, 2000).

Value

mse and the TSCAN clustering result.

Examples

data(Yan)
k = length(unique(trueclass))
Y = process_Y(Y, thre = 2) # preprocess the data
set.seed(123)
rixs = sample(nrow(Y), 500)
cixs = sample(ncol(Y), 40)
Y = Y[rixs, cixs]
con_res = Consensus(Y, k=k)
# not run
# mod_res = Select_Model_short_TSCAN(Y, cluster = con_res$cluster, top = c(100, 200))

suke18/FEAST documentation built on Sept. 14, 2021, 12:22 a.m.