generatecliques: Cliques generation and plot displaying.

In ubioinformat/TraRe: Transcriptional Rewiring

Cliques generation and plot displaying.

Description

By the way the LINKER method works, some highly-correlated driver genes (TFs) may be dropped from the resultant model, as the role they play at the GRN inference process is very similar. Due to this, we propose a method based on cliques (Fully Connected Networks) to recover those dropped drivers.

Usage

generatecliques(
  dataset = NULL,
  nassay = 1,
  method = "pearson",
  correlationth = 0.6,
  sparsecorrmatrix = TRUE,
  numcliques = "All",
  mandatorygenes = c(),
  selection = 1,
  table_cliques = TRUE
)

preparedata(dataset, method, table_cliques, correlationth)

generategraph(correlationth, sparsecorrmatrix, pdoutput)

selectmethod(selection, ggoutput, pdoutput)

plotcliques(ml, pdobject, sortparameter_ix_numcliques, smobject, numcliques)

Arguments

dataset	input expression file with genes as rows and samples as columns.
nassay	if SummarizedExperiment object is passed as input to dataset, name of the assay containing the desired matrix. Default: 0
method	method to use in the correlation matrix generation (see stats:cor). Default: 'pearson'
correlationth	threshold to consider edge exists. Default: 0.6
sparsecorrmatrix	boolean variable specifying whether to set to 0 values below threshold or not. Default: TRUE
numcliques	number of cliques to be generated. Default: 'All'
mandatorygenes	array of gene names which is mandatory to include in the returned cliques. Default: c() (none)
selection	integer selecting method. The available options are: 1 - Maximize Genes/Clique, 2 - Maximize Median Correlation Value/Clique, 3- Maximize Avg Variance Correlation Value/Clique or 4 - Maximize Sum(option two, option three).
table_cliques	boolean indicating to return all possible correlated genes instead of groups of cliques.
pdoutput	output from preparedata()
ggoutput	output from generategraph()
ml	cliques from smobject
pdobject	output from preparedata()
sortparameter_ix_numcliques	parameter for sorting cliques.
smobject	output from selectmethod()

Details

First, preparedata() prepares the correlation matrix and the hash table for future uses. Then, generategraph() generates an igraph object from genes correlation (threshold dependant) adjacency matrix. After that, selectmethod() chooses method in order to remove duplicities generated from the igraph::max_cliques method. Finally, generatecliques() generates all the cliques, containing fully connected networks of genes per clique.

Value

List containing the plot generated and a list with all the generated cliques.

Examples

   ## Suppose we want to recover the drivers LINKER may have dropped out.
   ## This method allows to group the highly correlated (above `correlationth`)
   ## driver genes so after the GRN generation, we know that if a particular
   ## driver gene is taking part of a relationship inside that GRN, all the
   ## genes inside this group may be also taking part of the same relationship
   ## due to the high correlation. Note that the large this threshold is, the
   ## surer we are about this affirmation.


   ## For this example, we will only work with the driver genes of a example dataset.

   dataset <- readRDS(paste0(system.file('extdata',package='TraRe'),
                      '/tfs_linker_example_eg.rds'))

   ## Lets select the generated dataset, as the rest of parameters are set by default.

   clioutput <- generatecliques(dataset = dataset)

ubioinformat/TraRe documentation built on March 10, 2024, 1:11 a.m.