R/identity_map.R

In elimereu/matchSCore2: What the Package Does TODOELI

#' Cell projections onto the reference dataset
#'
#' This function infers cell identities by using the model learned by the
#' reference dataset.
#'
#' @param scale.data A scaled matrix of gene expressions like in the `scale.data`
#' of the Seurat object. Rows are genes and columns are cells from the reference
#' dataset.
#' @param model The model as in output of the `train_model` function.
#' @param gene_cl.ref A list of cluster-specific markers. Each element of the list
#' contains markers of a cell type. Usually only top100 ranked markers are used.
#' @param p.threshold Probability threshold to consider a cell classified.
#' Default=0.5.
#'
#' @return A list with a vector with cell identities predicted by the model (ids)
#' and a dataframe with probabilities at each identity class.
#'
#' @export
#'
#' @examples
#' # TODO
identity_map <- function(scale.data,
                         model,
                         gene_cl.ref,
                         p.threshold = NULL) {
  p <- lapply(gene_cl.ref, function(x) rownames(scale.data)[which(rownames(scale.data) %in% x)])

  var.test <- sapply(p, function(x) Matrix::colSums(scale.data[x, ]))
  var.test <- t(apply(var.test, 1, function(x) (x - min(x)) / (max(x) - min(x))))

  model.test <- data.frame(var.test)
  fitted.results <- predict(model, newdata = model.test, "probs")

  fit <- apply(fitted.results, 1, function(x) colnames(fitted.results)[which(x == max(x))])
  if (is.null(p.threshold)) {
    p.threshold <- 0.5
  }
  fit <- apply(fitted.results, 1, function(x) ifelse(max(x) > p.threshold, colnames(fitted.results)[which(x == max(x))], "unclassified"))

  return(
    list(
      ids = fit,
      fit.prob = data.frame(fitted.results)
    )
  )
}