SCEnt (version 0.0.1)

scent_select: Feature Selection by Gene Heterogeneity

Description

Feature Selection by Gene Heterogeneity

Usage

scent_select(
  expr,
  bit_threshold = NULL,
  count_threshold = NULL,
  perc_threshold = NULL,
  unit = "log2",
  normalise = TRUE,
  transpose = FALSE
)

Arguments

expr

A matrix of gene expression data. Cells should be represented as rows and genes should be represented as columns.

bit_threshold

The threshold for the amount of bits of information a gene must add to be selected as a feature. Only one threshold can be used at a time.

count_threshold

A number represented how many of the most heterogeneous cells should be selected. Only one threshold can be used at a time.

perc_threshold

The percentile of the hetergeneity distribution above which a gene should be to be selected as a feature.

unit

The units to be used when calculating entropy.

normalise

A logical value representing whether the gene counts should be normalised into a probability distribution.

transpose

A logical value representing whether the matrix should be transposed before having any operations computed on it.

Value

A matrix of gene expression values where genes with low heterogeneity have been removed.

Examples

# NOT RUN {
#Creating Data
gene1 <- c(0,0,0,0,1,2,3)
gene2 <- c(5,5,3,2,0,0,0)
gene3 <- c(2,0,2,1,3,0,1)
gene4 <- c(3,3,3,3,3,3,3)
gene5 <- c(0,0,0,0,5,0,0)
gene_counts <- matrix(c(gene1,gene2,gene3,gene4,gene5), ncol = 5)
rownames(gene_counts) <- paste0("cell",1:7)
colnames(gene_counts) <- paste0("gene",1:5)

#Performing Feature Selection
scent_select(gene_counts, bit_threshold = 0.85)
scent_select(gene_counts, count_threshold = 2)
scent_select(gene_counts, perc_threshold = 0.25)
# }