MGstatistic: Statistics for ranking marker genes

In Lululuella/debCAM: Deconvolution by Convex Analysis of Mixtures

Description

This function computes One-Versus-Everyone Fold Change (OVE-FC) from subpopulation-specific expression profiles. Bootstrapping is optional.

Usage

MGstatistic(data, A = NULL, boot.alpha = NULL, nboot = 1000,
  cores = NULL)

Arguments

data	A data set that will be internally coerced into a matrix. Each row is a gene and each column is a sample. Data should be in non-log linear space with non-negative numerical values (i.e. >= 0). Missing values are not supported. All-zero rows will be removed internally.
A	When data are mixture expression profiles, A is estimated proportion matrix or prior proportion matrix. When data are pure expression profiles, A is a phenotype vector to indicate which subpopulation each sample belongs to.
boot.alpha	Alpha for bootstrapped OVE-FC confidence interval. The default is 0.05.
nboot	The number of boots.
cores	The number of system cores for parallel computing. If not provided, the default back-end is used.

Details

This function calculates OVE-FC and bootstrapped OVE-FC which can be used to identify markers from all genes.

Value

A data frame containing the following components:

idx	Numbers or phenotypes indicating which subpopulation each gene could be a marker for. If A is a proportion matrix without column name, numbers are returned. Otherwise, phenotypes.
OVE.FC	One-versus-Everyone fold change (OVE-FC)
OVE.FC.alpha	lower confidence bound of bootstrapped OVE-FC at alpha level.

Examples

#data are mixture expression profiles, A is proportion matrix
data(ratMix3)
MGstat <- MGstatistic(ratMix3$X, ratMix3$A)
## Not run: 
MGstat <- MGstatistic(ratMix3$X, ratMix3$A, boot.alpha = 0.05) #enable boot

## End(Not run)

#data are pure expression profiles without replicates
MGstat <- MGstatistic(ratMix3$S) #boot is not applicable
## Not run: 
#data are pure expression profiles with phenotypes
S <- matrix(rgamma(3000,0.1,0.1), 1000, 3)
S <- S[, c(1,1,1,2,2,2,3,3,3,3)] + rnorm(1000*10, 0, 0.5)
MGstat <- MGstatistic(S, c(1,1,1,2,2,2,3,3,3,3), boot.alpha = 0.05)

## End(Not run)

Lululuella/debCAM documentation built on May 14, 2021, 2:45 p.m.