R/cellClust.R

In GIS-SP-Group/RCA: RCA: Clustering Single Cell RNAseq Data using Reference Component Analysis

#' Generate cell clusters
#' 
#' Hierarchical clustering and dynamic tree cutting. 
#' 
#' @param obj_in data object.
#' @param method can only be "hclust" (default). 
#' @param deepSplit_wgcna integer value indicating how deep the dendrogram should be cut. Values can be 0, 1 (default), 2, 3 and 4. 
#' @param min_group_Size_wgcna integer value indicating the minimum size of the resulting clusters. Default is 5.
#' @return data object.
#' @export
#' @examples
#' 
#' data_obj = cellClust(data_obj);
#' 
cellClust <- function(obj_in,method="hclust",deepSplit_wgcna=1,min_group_Size_wgcna=5)
{
  #### 1: reading input ####  
  fpkm_temp = obj_in$fpkm_for_clust;
  
  if (!require(flashClust)) install.packages("flashClust",repos = "http://cran.us.r-project.org") 
  require(flashClust)
  if (!require(WGCNA)){
    source("http://bioconductor.org/biocLite.R");
    biocLite(c("impute", "GO.db", "preprocessCore")); 
    install.packages("WGCNA");
  }
  require(WGCNA)  
  
  #### 2: choosing method ####
    if (method == "hclust"){
        d = as.dist(1-cor(fpkm_temp,method="pearson"));
        cellTree = flashClust(d,method = "average");     
        dynamicGroups = cutreeDynamic(dendro = cellTree,distM = as.matrix(d),deepSplit = deepSplit_wgcna,
                                      pamStage = FALSE,minClusterSize= min_group_Size_wgcna);
        dynamicColors = labels2colors(dynamicGroups);
        group_labels = matrix(dynamicGroups,nrow = length(names(fpkm_temp)),ncol = 1,list(names(fpkm_temp),c("groupLabel")),byrow=FALSE)
        group_labels = as.data.frame(group_labels)
        group_labels_color = cbind(group_labels,dynamicColors);
    }
  #### 3: writing output ####    
  obj_out = append(obj_in,
              list("d" = d,
              "cellTree" = cellTree,
              "dynamicColors" = dynamicColors,
              "group_labels_color" = group_labels_color
              )
  )
  
  return(obj_out)
  
}