correlation {MetNet}R Documentation

Create an adjacency matrix based on correlation

Description

correlation infers an adjacency matrix using correlation using the corAndPvalue function (from the WGCNA package), pcor (from ppcor) or spcor (from ppcor). correlation extracts the reported p-values from the function corAndPvalue, pcor or spcor that can be adjusted for multiple testing (correlation_adjust parameter) and will return an unweighted adjacency matrix containing edges if the (adjusted) p-value is below the value defined by correlation_threshold.

Usage

correlation(x, correlation_adjust="none", type="pearson", 
                                        correlation_threshold=0.05, ...)

Arguments

x

matrix, where columns are the samples and the rows are features (metabolites), cell entries are intensity values

correlation_adjust

character

type

character, either "pearson", "spearman", "pearson_partial", "spearman_partial", "pearson_semipartial" or "spearman_semipartial". type will be passed to argument method in corAndPvalue (in the case of "pearson" or "spearman") or to method in pcor ("pearson" and "spearman" for "pearson_partial" and "spearman_partial", respectively) or to method in spcor ("pearson" or "spearman" for "pearson_semipartial" and "spearman_semipartial", respectively)

correlation_threshold

numeric, significance level α (default: 0.05), if the (adjusted) p-values exceed this value, there is no statistical connection between features

...

parameters passed to corAndPvalue (argument adjust will be ignored)

Details

If "pearson" or "spearman" is used as a method the function corAndPvalue from WGCNA will be employed. If "pearson_partial" or "spearman_partial" is used as a method the function pcor from spcor will be employed. If "pearson_semipartial" or "spearman_semipartial" is used as a method the function spcor from spcor will be employed. For use of the parameters used in the corAndPvalue function, refer to ?WGCNA::corAndPvalue.

Value

matrix, matrix with edges inferred from correlation algorithm corAndPvalue, pcor or spcor (depending on the chosen method)

Author(s)

Thomas Naake, thomasnaake@googlemail.com

Examples

data("x_test", package="MetNet")
x <- x_test[, 3:dim(x_test)[2]]
x <- as.matrix(x)
correlation(x, correlation_adjust="bonferroni", type="pearson")

[Package MetNet version 1.0.1 Index]