Group-centralized principal component analysis

Group-centralized principal component analysis

Usage

Run.gcPCA(
  X,
  meta,
  g_factor,
  b_factor,
  lambda = 50,
  npcs = 30,
  excluded.cells = NULL,
  output.all = FALSE,
  do.scale = F,
  do.cosine = F,
  block.sizeGB = 2,
  block.size = 4000
)

Arguments

X

Input data matrix.

meta

Input metadata (data.frame).

g_factor

Group variable (s).

b_factor

Batch variable (s).

lambda

Tuning parameter.

npcs

How many dimensions to reduce.

excluded.cells

Metadata about which cells should not be used.

output.all

Whether to return all calculated values.

do.scale

Whether to perform global scaling.

do.cosine

Whether to cosine normalize the integration result.

block.sizeGB

Memory settings for block computing.

block.size

Block size.