SHARP_large: Run SHARP for large-size (by default, >= 5000) single-cell...

In shibiaowan/SHARP: hyperfast and accurate single-cell RNA-Seq data clustering

Description

For large-size (>= 5000) datasets, we suggest first partitioning the datasets into several groups, then we run SHARP for each group, and finally and we ensemble the results of each group by a similarity-based meta-clustering algorithm.

Usage

SHARP_large(
  scExp,
  ncells,
  ensize.K,
  reduced.dim,
  partition.ncells,
  hmethod,
  N.cluster,
  enpN.cluster,
  indN.cluster,
  minN.cluster,
  maxN.cluster,
  sil.thre,
  height.Ntimes,
  flashmark,
  flag,
  n.cores,
  forview,
  rM,
  rN.seed
)

Arguments

scExp	input single-cell expression matrix
ncells	number of single cells
ensize.K	number of applications of random projection for ensemble
reduced.dim	the dimension to be reduced to
partition.ncells	number of cells for each partition when using SHARP_large

Details

For each partition (or group), the default number of cells is set to 2000 for each group. The users can also set a different number according to the computational capability of their own local computers. The suggested criteria to set this number is that as long as SHARP_small can run in a fast enough (depending on users' requirements) way for the selected number of single cells.

Examples

enresults = SHARP_large(scExp, ncells, ensize.K, reduced.dim, partition.ncells)

shibiaowan/SHARP documentation built on April 28, 2021, 1:56 p.m.