R/phenograph_hm.R

In douve/viRievac: Main R functions (viRievac)

# Create a Heatnap from Rphenograph results:
# data must have a variable named 'louvain', the output from Rphenograph()


phenograph_hm = function(data,
                         group.var,
                         vars,
                         color=colorRampPalette(c("blue4","#0571B0","white","red1","#CA0020"))(100),
                         statistic='median'){

  aux = data %>% dplyr::select(group.var,vars) %>%
    group_by_(group.var)
  if(statistic=='median') aux = aux %>% dplyr::summarise_all(median) %>% select(-group.var) %>% t() %>% data.frame
  if(statistic=='mean') aux = aux %>% dplyr::summarise_all(mean) %>% select(-group.var) %>% t() %>% data.frame
  colnames(aux) = table(data[,group.var]) %>% names

  # Calculate cluster frequencies
  clustering_table <- as.numeric(table(data %>% select(group.var)))
  labels_col <- paste0(colnames(aux), " (",
                       round(clustering_table / sum(clustering_table) * 100, 2), "%)")

  # heatmap parameters
  hm.parameters <- list(aux,
                        color = color,
                        scale = "row",
                        breaks = seq(-3.5,3.5,7/100),
                        cellwidth =20,cellheight=15,
                        kmeans_k = NA,
                        show_rownames = T, show_colnames = T,
                        main = "",
                        clustering_method = "ward.D2",
                        cluster_rows = T,
                        cluster_cols = T,
                        clustering_distance_rows = "euclidean",
                        clustering_distance_cols = "euclidean",
                        labels_col = labels_col,
                        display_numbers = TRUE, number_color = "black",
                        fontsize = 8,
                        fontsize_number = 0
  )

  do.call('pheatmap',hm.parameters)

}