ssc.reduceDim: Reduce dimension by various methods

In Japrin/sscClust: simpler single cell RNAseq data clustering

Reduce dimension by various methods

Description

Reduce dimension by various methods

Usage

ssc.reduceDim(
  obj,
  assay.name = "exprs",
  method = "iCor",
  method.vgene = "HVG.sd",
  method.tsne = "Rtsne",
  pca.npc = NULL,
  tSNE.usePCA = T,
  tSNE.perplexity = 30,
  autoTSNE = T,
  dim.name = NULL,
  iCor.niter = 1,
  iCor.method = "spearman",
  reuse = F,
  ncore = NULL,
  seed = NULL,
  out.prefix = NULL,
  ...
)

Arguments

obj	object of singleCellExperiment class
assay.name	character; which assay (default: "exprs")
method	character; method to be used for dimension reduction, should be one of (pca, tsne, iCor). (default: "iCor")
method.vgene	character; method to identify variable genes. (default: sd)
method.tsne	character; method to run tsne, one of "Rtsne", "FIt-SNE". (default: "Rtsne")
pca.npc	integer; number of pc be used. Only for reduction method "pca". (default: NULL)
tSNE.usePCA	logical; whether use PCA before tSNE. Only for reduction method "tsne". (default: T)
tSNE.perplexity	logical; perplexity parameter. Used in all Rtsne() calling. (default: 30)
autoTSNE	logical; Wheter generate automatically a tSNE map when reduction method is "pca" or "iCor". (default: T)
dim.name	character; store the reduced data under the name in the obj's reducedDim SimpleList. If i it is NULL, infer from method. (default: NULL)
iCor.niter	integer; number of iteration of calculating the correlation. Used in reduction method "iCor". (default: 1)
iCor.method	character; method to calculate correlation between samples, should be one of "spearman" and "pearson". (default "spearman")
reuse	logical; don't calculate if the query is already available. (default: F)
ncore	integer; nuber of CPU cores to use. if NULL, automatically detect the number. (default: NULL)
seed	integer; seed of random number generation. (default: NULL)
out.prefix	character; output prefix (default: NULL)
...	parameters passed to RD methods

Details

If the reduction method is "pca", the function will call prcomp() and estimate the number of top PC should be used in downstream analysis using and "elbow" based method, then the samples coordinates in the space spaned by the top PC would stored in the reducedDim slot of the return value with the reducedDimName "pca".If autoTSNE is 'true', a tSNE map based on the top PC will generated and stored in the reducedDim slot with reducedDimName "pca.tsne".

If the reduction method is "tsne", a tSNE map will generated and stored in the reducedDim slot with reducedDimName "tsne". If tSNE.usePCA is 'true', maximum 50 top PC will be used, but no estimation based on "elbow" method performed.

If the reduction method is "iCor", the function will calculate correlation between samples for iCor.niter times. If autoTSNE is 'true', it also generate automatically a tSNE map based on the correlations. These data would be stored in the reducedDim slot with reducedDimName "iCor" and "iCor.tsne" respectively.

Value

an object of SingleCellExperiment class with reduced data added to the reducedDim slot.

Japrin/sscClust documentation built on May 23, 2024, 7:12 a.m.