perm.significance {MVisAGe}

A Function for Computing a Vector of Pearson Correlation Coefficients

Description

This function computes Pearson correlation coefficients on a row-by-row basis for two numerical input matrices of the same size.

Usage

perm.significance(exp.mat, cn.mat, gene.annot, method = "pearson",
  digits = 5, num.perms = 100, random.seed = NULL,
  alternative = "greater")

Arguments

exp.mat

A matrix of gene-level expression data (rows = genes, columns = samples). Missing values are not permitted.

cn.mat

A matrix of gene-level DNA copy number data (rows = genes, columns = samples). Both genes and samples should appear in the same order as exp.mat. Missing values are not permitted.

gene.annot

A three-column matrix containing gene position information. Column 1 = chromosome number written in the form 'chr1' (note that chrX and chrY should be written chr23 and chr24), Column 2 = position (in base pairs), Column 3 = cytoband. Genes should appear in the same order as exp.mat and cn.mat.

method

A character string (either "pearson" or "spearman") specifying the method used to calculate the correlation coefficient (default = "pearson").

digits

Used with signif() to specify the number of significant digits (default = 5).

num.perms

Number of permutations used to assess significance (default = 1e2).

random.seed

Random seed (default = NULL).

alternative

A character string ("greater" or "less") that specifies the direction of the alternative hypothesis, either rho > 0 or rho < 0 (default = "greater").

Value

Returns a five-column matrix. The first three columns are the same as gene.annot. The fourth column contains gene-specific Pearson or Spearman correlation coefficients based on the entries in each row of exp.mat and cn.mat, respectively (column name = "R"). The fifth column contains squared Pearson correlation coefficients (column name = "R^2"). The sixth column contains the permutation-based right-tailed p-value of the correlation coefficient (column name = "perm_pValue"). The seventh column contains Benjamini-Hochberg q-values corresponding to the p-values. Genes with constant gene expression or DNA copy number are removed because they have zero variance.

Examples

exp.mat = tcga.exp.convert(exp.mat)

 cn.mat = tcga.cn.convert(cn.mat)

 prepped.data = data.prep(exp.mat, cn.mat, gene.annot, sample.annot, log.exp = FALSE)

 perm.significance(prepped.data[["exp"]], prepped.data[["cn"]], prepped.data[["gene.annot"]])