identifyModules: Module Identification
In krumsieklab/MoDentify: Phenotype-driven module identification

Description Usage Arguments Value References Examples

Module Identification

identifyModules(graph, data, phenotype, covars = NULL, annotations,
merge.overlapping=FALSE, better.than.components= TRUE, alpha=0.05,
level=NULL, representative.method="average", correction.method="bonferroni",
BPPARAM = SerialParam(progressbar = TRUE))

`graph`	an `igraph` object, which can be generated with `generateNetwork`. The ID of the nodes must correspond to the name of the variables.
`data`	either a matrix, where the columns correspond to the variables and the rows to the observations. Or a `data.table` with three columns: name, sampleID and value.
`phenotype`	a vector with the values for a phenotype of interest. It must have the same number of observations as data.
`covars`	a `data.table` containing the covariates to correct for. The rows for the samples must be in the same order as in the phenotype vector.
`annotations`	a `data.table` containing annotations for the variables. The columns correspond to the different annotations, the rows to the variables.
`merge.overlapping`	if `TRUE`, overlapping modules will be merged.
`better.than.components`	if `TRUE`, modules will only be enlarged and accepted, if they are better than all of their components.
`alpha`	significance level for accepting the modules.
`level`	Must be set to the name of the column to be used, if modules should be calculated for pathways.
`representative.method`	the method, that is used for the calculation of the module representation. Currently implemented: "eigenmetabolite" and "average"
`correction.method`	the method that used for multiple testing correction ("bonferroni", "BH", "BY", "fdr", "holm", "hochberg", "hommel", "none"). Default is set to bonferroni. See `p.adjust`.
`BPPARAM`	An instance of the `BiocParallelParam-class` that determines how to parallelisation of the functions will be evaluated.
`scoringFunction`	a scoring function accepting parameters moduleRepresentatives, phenotype and covars. See `linearScoring`

a list consisting of four elements.

\insertRef

Do2017MoDentify

\insertRef

Chuang2007MoDentify

data(qmdiab.data)
data(qmdiab.annos)
data(qmdiab.phenos)

data <- qmdiab.data[, 1:75]
annotations <- qmdiab.annos[1:75]

net.graph <- generateNetwork(data = data, annotations = annotations)
mods <- identifyModules(
  graph = net.graph, data = data, annotations =
    annotations, phenotype = qmdiab.phenos$T2D, alpha = 0.05
)

pathway.graph <- generatePathwaysNetwork(data = data, annotations = annotations)

pathway.modules <- identifyModules(
  graph = pathway.graph$network, data = data,
  phenotype = qmdiab.phenos$T2D, level = pathway.graph$level, annotations = annotations,
  alpha = 0.05
)