R/Functions-fromSimilarity.R

In milescsmith/WGCNA: Weighted Correlation Network Analysis

# Functions to perform WGCNA from similarity input.

matrixToNetwork <- function(mat,
                            symmetrizeMethod = c("average", "min", "max"),
                            signed = TRUE,
                            min = NULL,
                            max = NULL,
                            power = 12,
                            diagEntry = 1) {
  sm <- match.arg(symmetrizeMethod)
  if (is.na(sm)) {
    stop("Unrecognized or non-unique 'symmetrizeMethod'.")
  }

  mat <- as.matrix(mat)

  nd <- 0
  x <- try({
    nd <- dim(mat)
  })
  if ((class(x) == "try-error") | (nd != 2)) {
    stop("'mat' appears to have incorrect type; must be a 2-dimensional square matrix.")
  }

  if (ncol(mat) != nrow(mat)) {
    stop("'mat' must be a square matrix.")
  }

  if (!signed) mat <- abs(mat)

  if (sm == 1) {
    mat <- (mat + t(mat)) / 2
  } else if (sm == 2) {
    mat <- pmin(mat, t(mat), na.rm = TRUE)
  } else {
    mat <- pmax(mat, t(mat), na.rm = TRUE)
  }

  if (is.null(min)) {
    min <- min(mat, na.rm = TRUE)
  } else {
    mat[mat < min] <- min
  }

  if (is.null(max)) {
    max <- max(mat, na.rm = TRUE)
  } else {
    mat[mat > max] <- max
  }

  adj <- ((mat - min) / (max - min))^power

  diag(adj) <- diagEntry

  adj
}

checkSimilarity <- function(similarity, min = -1, max = 1) {
  checkAdjMat(similarity, min, max)
}

adjacency.fromSimilarity <- function(similarity, type = "unsigned", power = if (type == "distance") 1 else 6) {
  checkSimilarity(similarity)
  adjacency(similarity,
    type = type, power = power, corFnc = "I", corOptions = "", distFnc = "I",
    distOptions = ""
  )
}

softConnectivity.fromSimilarity <- function(similarity, type = "unsigned",
                                            power = if (type == "signed") 15 else 6,
                                            blockSize = 1500, verbose = 2, indent = 0) {
  checkSimilarity(similarity)
  softConnectivity(similarity,
    corFnc = "I", corOptions = "",
    type = type, power = power,
    blockSize = blockSize, verbose = verbose, indent = indent
  )
}

pickHardThreshold.fromSimilarity <- function(similarity,
                                             RsquaredCut = 0.85, cutVector = seq(0.1, 0.9, by = 0.05),
                                             moreNetworkConcepts = FALSE, removeFirst = FALSE, nBreaks = 10) {
  checkSimilarity(similarity)
  pickHardThreshold(similarity,
    dataIsExpr = FALSE, RsquaredCut = RsquaredCut, cutVector = cutVector,
    moreNetworkConcepts = moreNetworkConcepts, removeFirst = removeFirst,
    nBreaks = nBreaks, corFnc = "I", corOptions = ""
  )
}

pickSoftThreshold.fromSimilarity <- function(similarity,
                                             RsquaredCut = 0.85, powerVector = c(seq(1, 10, by = 1), seq(12, 20, by = 2)),
                                             removeFirst = FALSE, nBreaks = 10, blockSize = 1000,
                                             moreNetworkConcepts = FALSE, verbose = 0, indent = 0) {
  checkSimilarity(similarity)
  pickSoftThreshold(similarity,
    dataIsExpr = FALSE, RsquaredCut = RsquaredCut, powerVector = powerVector,
    removeFirst = removeFirst, nBreaks = nBreaks,
    blockSize = blockSize, networkType = "signed",
    moreNetworkConcepts = moreNetworkConcepts,
    verbose = verbose, indent = indent
  )
}