dpm.R

#' Function to compute DPM
#'
#' @param data is a matrix containing samples in rows, variables in columns
#'
#' @return a nvar*nvar matrix containing reg.DPM scores for all links (edges) where nvar is the number of variables
#'
#' @import corpcor
#'
#' @examples
#' Example1:
#' ## Generate non-linear data with 100 samples (from the network discussed in the paper) where the gaussian noise with standard deviation of sigma=0.2 is added to the generated data
#' data <- nldata(nsamp=100,sigma=0.2)$data
#'
#'## run regularized DPM
#' res <-reg.dpm(data)
#' head(res)
#'
#' Example2:
#' ##load DREAM4 data (insilico_size10_1_knockouts)
#' data(dream4)
#'
#'## Run regularized DPM
#' res <- reg.dpm(dream4)
#' head(res)
#'
#' @export

###############################
### Function to compute DPM ###
###############################

dpm <- function(data) {
  # compute the double-centred vectors of each variable and concatenate them into the columns of matrix d
  d <- apply(as.matrix(data), 2, Dcenter)
  # compute the precision matrix of d (cor2pcor is a function from corpcor Library)
  res <- corpcor::cor2pcor(cov(d))
  return(res)
}