Skip to contents

CNV analysis

Usage

CNV_analysis(
  scMuffinList = NULL,
  mc.cores = 1,
  reference = NULL,
  min_cells = 100,
  min_genes = 200,
  wnd_size = 100,
  center_cells = TRUE,
  center_genes = FALSE,
  expr_lim = FALSE,
  method = "mean",
  na.rm = FALSE,
  z.score = FALSE,
  eps = NULL,
  n_comp = 10,
  gene_ann = NULL,
  ...
)

Arguments

scMuffinList

scMuffinList object

mc.cores

number of cores;

reference

one-column data.frame with a reference expression profile; rownames must match those of genes_by_cell;

min_cells

minimum numbver of cells in which a gene must be expressed;

min_genes

minimun number of genes expressed in a cell;

wnd_size

number of adjacent genes considered;

center_cells

whether to center cells. Default is TRUE

center_genes

whether to center genes or not. Default is FALSE. When using a reference it is useful to set this to TRUE

expr_lim

min and max values of relative expression; by default, lower and higher whiskers returned by grDevices::boxplot.stats will be used. Set to NA or anything other value such that length(expr_lim) != 2 to disable the removal of outliers.

method

"mean": subtract the average profile of the reference cluster to every cell (default); "min_max" (from Tirosh et al., DOI: 10.1126/science.aad0501) can be used but is not tested yet

na.rm

whether to remove 0 values in CNV estimation

z.score

whether to use z-scores of the cluster median CNV profile, instead of the median itself.

eps

absolute threshold to call CNV regions.

n_comp

number of PCA components used for clustering

gene_ann

optional data.frame with gene annotation with mandatory columns "Chromosome", "symbol" and "pos" (genomic location). If NULL gene locations will be collected from org.Hs.eg.db

...

arguments passed to Seurat::FindClusters()