statistical: Create a list of adjacency matrices from statistical methods

In MetNet: Inferring metabolic networks from untargeted high-resolution mass spectrometry data

Description

The function 'statitical' infers adjacency matrix topologies from statistical methods and returns matrices of these networks in a 'list'. The function includes functionality to calculate adjacency matrices based on LASSO (L1 norm)-regression, random forests, context likelihood of relatedness (CLR), the algorithm for the reconstruction of accurate cellular networks (ARACNE), Pearson correlation (also partial and semipartial), Spearman correlation (also partial and semipartial) and score-based structure learning (Bayes). The function returns a list of adjacency matrices that are defined by 'model'.

Usage

statistical(x, model, ...)

Arguments

x	'matrix' that contains intensity values of features/metabolites (rows) per sample (columns).
model	'character' vector containing the methods that will be used ('"lasso"', '"randomForest"', '"clr"', '"aracne"', '"pearson"', '"pearson_partial"', '"pearson_semipartial"', '"spearman"', '"spearman_partial"', '"spearman_semipartial"', '"bayes"')
...	parameters passed to the functions 'lasso', 'randomForest', 'clr', 'aracne', 'correlation' and/or 'bayes'

Details

The function 'statistical' includes functionality to calculate adjacency matrices based on LASSO (L1 norm)-regression, random forests, context likelihood of relatedness (CLR), the algorithm for the reconstruction of accurate cellular networks (ARACNE), Pearson correlation (also partial and semipartial), Spearman correlation (also partial and semipartial) and Constraint-based structure learning (Bayes).

'statistical' calls the function 'lasso', 'randomForest', 'clr', 'aracne', 'correlation' (for '"pearson"', '"pearson_partial"', '"pearson_semipartial"', '"spearman"', '"spearman_partial"', '"spearman_semipartial"') and/or 'bayes' as specified by 'model'. It will create adjacency matrices using the specified methods and will return a 'list' containing the weighted adjacency matrices.

Internally 'x' will be z-scaled and the z-scaled object will be used in 'lasso', 'clr' and/or 'aracne'.

Value

'list' containing the respective adjacency matrices specified by 'model'

Author(s)

Thomas Naake, [e-mail]

Examples

data("x_test", package = "MetNet")
x <- x_test[1:10, 3:ncol(x_test)]
x <- as.matrix(x)
statistical(x = x, model = c("pearson", "spearman"))

MetNet documentation built on Nov. 8, 2020, 7:34 p.m.