Skip to contents

FindVariableFeaturesMix

FindVariableFeaturesMix.Rd

FindVariableFeaturesMix

Usage

FindVariableFeaturesMix(
  object,
  method.names = c("scran", "scran_pos", "seuratv1"),
  nfeatures = 2000,
  loess.span = 0.3,
  clip.max = "auto",
  num.bin = 20,
  binning.method = "equal_width",
  verbose = FALSE
)

Arguments

object

An object, SeuratObject and matrix(including sparse matrix) are both acceptable

method.names

The following methods can be directly used for highly variable feature selection. The mixture of methods take a vector of method list, e.g. c("scran","scran_pos","seuratv1"), which is also default.

  • scran: Use mean-variance curve adjustment on lognormalized count matrix, which is scran ModelGeneVar.

  • mv_ct: Use mean-variance curve adjustment on count matrix, inherited from scran ModelGeneVar.

  • mv_nc: Use mean-variance curve adjustment on normalized count matrix, inherited from scran ModelGeneVar.

  • mv_lognc: The same as scran.

  • mv_PFlogPF: Use mean-variance curve adjustment on PFlog1pPF matrix, inherited from scran ModelGeneVar.

  • scran_pos: Use scran poisson version, modelGeneVarByPoisson.

  • seuratv3: Use logmean-logvariance curve adjustment on count matrix, which is vst, Seurat FindVariableFeatures Function(https://satijalab.org/seurat/reference/findvariablefeatures).

  • logmv_ct: The same as seuratv3.

  • logmv_nc: Use logmean-logvariance curve adjustment on normalized count matrix, inherited from seuratv3(vst).

  • logmv_lognc: Use logmean-logvariance curve adjustment on lognormalized count matrix, inherited from seuratv3(vst).

  • logmv_PFlogPF: Use logmean-logvariance curve adjustment on PFlog1pPF matrix, inherited from seuratv3(vst).

  • seuratv1: Use dispersion on lognormalized count matrix, which is dispersion (disp), Seurat FindVariableFeatures Function(https://satijalab.org/seurat/reference/findvariablefeatures).

  • disp_lognc: The same as seuratv1.

  • disp_PFlogPF: Use dispersion on PFlog1pPF matrix, inherited from seuratv1(disp).

  • mean_max_ct: Highly Expressed Features with respect to count matrix.

  • mean_max_nc: Highly Expressed Features with respect to normalized count matrix.

  • mean_max_lognc: Highly Expressed Features with respect to lognormalized count matrix

nfeatures

Number of features to select as top variable features.

loess.span

(Only work for logmv based methods like seuratv3). Loess span parameter used when fitting the variance-mean relationship

clip.max

(Only work for logmv based methods like seuratv3). After standardization values larger than clip.max will be set to clip.max; default is 'auto' which sets this value to the square root of the number of cells

num.bin

(Only work for logmv or dispersion based methods)Total number of bins to use in the scaled analysis (default is 20)

binning.method

Specifies how the bins should be computed. Available methods are:

  • equal_width: each bin is of equal width along the x-axis[default].

  • equal_frequency: each bin contains an equal number of features (can increase statistical power to detect overdispersed features at high expression values, at the cost of reduced resolution along the x-axis).

verbose

Whether to show progress bar for calculations. Default is FALSE.

Value

object: If the input is SeuratObject, the return is also SeuratObject; if the input is matrix(including sparse matrix), the return is the highly variable feature names.

Details

The function inherits from FindVariableFeatures function of Seurat Package. Refer to https://github.com/RuzhangZhao/mixhvg for user manual.

Examples

if(0){
simple_matrix<-matrix(1:2e4,nrow=4000,ncol=5)
rownames(simple_matrix)<-1:nrow(simple_matrix)
colnames(simple_matrix)<-1:ncol(simple_matrix)
simple_matrix_HVG<-FindVariableFeaturesMix(simple_matrix)
}