buildExpNet: Build experimental networks
In MetClassNet/MetClassNetR: Create and analize networks built from metabolomics Data

buildExpNet

R Documentation

Build experimental networks

Description

The function buildExpNet builds experimental networks, based on spectral similarity, correlation, and mass difference

Usage

buildExpNet(
  inputData,
  net2Build = "all",
  directed = FALSE,
  ppmMass = 10,
  ppmSpec = 0,
  tol = 0.005,
  corrModel = "pearson",
  corrThresh = 0.25,
  specSimThresh = 0.7
)

Arguments

`inputData`	`list`, list returned by the `loadInputData` function
`net2Build`	`list`, list of experimental networks to build, according to the following options: "all" - generates 3 experimental networks: mass difference, correlation, and spectral similarity, "m" - builds the mass difference network, "s" - builds the spectral similarity network, "c" - builds the correlation network. It is to note that combinations of m, s, and c are possible, writing them separated by commas, e.g., c("m", "c") would generate the mass difference, and correlation networks. The default value is "all"
`directed`	`boolean`, boolean value (TRUE/FALSE). If TRUE then the networks that are generated will be directed, and undirected otherwise. The default value is FALSE (i.e., undirected network)
`ppmMass`	`numeric`, (optional) allowed error for mass differences calculus. It is only needed if the mass difference network is to be built. The default value is 10
`ppmSpec`	`numeric`, (optional) relative allowed error for spectral similarity calculus. It is only needed if the spectral similarity network is to be built. The default value is 0
`tol`	`numeric`, (optional) absolute tolerance for spectral similarity calculus. It is only needed if the spectral similarity network is to be built. The default value is 0.005
`corrModel`	`vector`, (optional) character vector containing the model(s) to be used for the correlation calculus in MetNet. Please check the available models in the current version of MetNet. There are 10 models available in version 1.14.0: "lasso", "randomForest", "clr", "aracne", "pearson", "pearson_partial", "spearman", "spearman_partial", "ggm", and "bayes". This parameter is only needed if the correlation network is to be built. The default value is "pearson"
`corrThresh`	`numeric`, (optional) floating point number indicating the correlation threshold to consider that two features are correlated. It is only needed if the correlation network is to be built. The default value is 0.25 (i.e., at least 25% of correlation between the abundance values, either positive or negative)
`specSimThresh`	`numeric`, (optional) floating point number indicating the spectral similarity threshold to consider that two features have similar spectra. It is only needed if the spectral similarity network is to be built. The default value is 0.7