IGST (version 0.1.0)

IGST.pval.bootmrmrsvm: Computation of statistical significance values for genes using SVM and MRMR feature selection technique with bootstrapping procedure

Description

The function computes the statistical significance values for the genes from the non-parametric test "H0: i-th gene is not informative against H1: i-th gene is informative" for selection of informative genes using SVM and MRMR feature selection technique with bootstrapping procedure.

Usage

IGST.pval.bootmrmrsvm(x, y, re, Q, v)

Arguments

x

x is a n by p data frame of gene expression values where rows represent genes and columns represent samples. Each cell entry represents the expression level of a gene in a sample or subject (row names of x as gene names or gene ids).

y

y is a p by 1 numeric vector with entries 1 or -1 representing sample labels, where, 1 or -1 represents the sample label of subjects or samples for stress or control condition(for two class problems).

Q

Q is a scalar representing the quartile value of the rank scores of genes (lies within 1\/N to 1), usually the second quartile, i.e. 0.5 or third quartile i.e. 0.75 may be taken.

v

v is a scalar representing the weightage of a method and must be within 0 and 1.

re

re is a scalar representing the number of bootstrap generated, re must be sufficiently large (i.e. number of times bootstrap samples are generated.

Value

The function returns a vector of p-values for all the genes from the given statistical test in the dataset using SVM and MRMR feature selection technique with bootstrapping procedure.

References

Das, S., Rai, A., Mishra, D. C., & Rai, S. N. (2018). Statistical approach for selection of biologically informative genes. Gene, 655, 71-83.

Examples

# NOT RUN {
#################################
library(IGST)
data(rice_cold)
x<-rice_cold[-1,]
y<-rice_cold[1,]
y<-as.matrix(y)
y<-as.vector(y)
#s<-10
Q<-0.5
v<-0.5
re<-10
IGST.pval.bootmrmrsvm (x, y, re, Q, v)
# }