0

我正在使用 ComplexHeatmap 中的 oncoprint() 函数，并且一直看到此警告：

“您cell_fun为超过 100 行或列的热图定义，绘制起来可能非常慢。考虑使用矢量化版本 layer_fun。”

根据作者的指南，cell_fun 和 layer_fun 与同一包中的 Heatmap() 函数相关（章节链接）。该指南没有提及关于 oncoprint() 的 layer_fun。

我浏览了 github，我看到当输入矩阵大于 100x100 时会发生此错误。我看到 layer_fun 也以某种方式参与了 oncoprint() 函数，但我不明白它是如何参与的文档在 oncoprint 中提到 layer_fun。显然，它与图形表示“函数列表”变量 alter_fun 相关联。

我不明白在 oncoprint() 函数中 alter_fun 和 layer_fun 是如何相互关联的。但是，我使用了热图注释（参见变量 ha_all）——这是否会导致 layer_fun 警告（我不这么认为）。

以下是我的问题：矢量化 alter_fun 是什么意思，我该怎么做？layer_fun 与这一切有关吗？将来，我的输入矩阵会更大，所以为了速度，我需要对这些变量进行向量化。

# Toy Example, Big Data ---------------------------------------

# Colors for Each Annotation (also used to randomly create toy data)

toy_gender <- c("Male" = "blue", "Female" = "pink")

toy_grade <- c("1" = "white", "2" = "gray", "3" = "dark gray", "4"="black")

toy_mol <- c("Glioma " = "yellow",
              "Meningioma" = "black",
              "Ependymoma" = "pink")

toy_col <-  # alteration type
  c(
    "arm_level_gain" = "blue",
    "wt" = "beige",
    "missense" = "pink",
    "arm_level_loss" = "red",
    "damaging_mutation" = "purple"
  )


toym <-  #150 x 150
  data.frame("genes" = c(1:150))

the_cols <- # each sample will be a column in matrix
  rep("sample", 150) %>% paste(., c(1:150), sep="")


toym$genes <- paste("g",  toym$genes, sep="")


for (i in the_cols){  #create cols: each is one sample
  # fill with random alts data 
  toym[, i] <- sample( (names(toy_col)) , 150, replace = TRUE)

  }

#label rows as gene name, then rm sample col
row.names(toym)<- toym$genes

toym <- toym[,2:length(toym)]

toym <- as.matrix(toym)


# Create Annotation Data Frame
toydf <- data.frame( "Samples" = colnames(toym) )

toydf$Molecular <- sample( names(toy_mol), 150, replace=TRUE )

toydf$Gender <- sample( names(toy_gender), 150, replace=TRUE )

toydf$Grade <- sample( names(toy_grade), 150, replace=TRUE )

row.names(toydf) <- toydf$Samples

toydf <- toydf[, 2:4]


ha_all = HeatmapAnnotation(  #constructor function-- read complexheatmap documentation
  df = toydf,
  col = list(Molecular = toy_mol,
             Grade = toy_grade,
             Gender = toy_gender),
  na_col = "grey"
)


# Create alter_fun list of functions
toy_alter_fun = list(
  
  background = function(x, y, w, h) {
    grid.rect(x, y, w-unit(0.5, "mm"), h-unit(0.5, "mm"),
              gp = gpar(fill = "beige", col = NA))
  },
  
  "focal_gain" = function(x, y, w, h) {
    grid.rect(x, y, w-unit(0.5, "mm"), h-unit(0.5, "mm"),
              gp = gpar(fill = toy_col["focal_gain"], col = NA))
  },
  
  "damaging_mutation" = function(x, y, w, h) {
    grid.rect(x, y, w-unit(0.5, "mm"), h-unit(0.5, "mm"),
              gp = gpar(fill = toy_col["damaging_mutation"], col = NA))
  },
  
  # big blue
  #"wt" = function(x, y, w, h) {
  #   grid.rect(x, y, w-unit(0.5, "mm"), h-unit(0.5, "mm"),
  #      gp = gpar(fill = "beige", col = NA))
  #},
  
  "missense" = function(x, y, w, h) {
    grid.rect(x, y, w-unit(0.5, "mm"), h-unit(0.5, "mm"),
              gp = gpar(fill = toy_col["missense"], col = NA))
  },
  
  
  # bug red
  
  "rearrangement" = function(x, y, w, h) {
    grid.rect(x, y, w-unit(0.5, "mm"), h-unit(0.5, "mm"),
              gp = gpar(fill = toy_col["rearrangement"], col = NA))
  },
  
  # # small green
  
  "focal_loss" = function(x, y, w, h) {grid.rect(x, y, w*0.9, h*0.4,
                                                 gp = gpar(fill = toy_col["focal_loss"], col = NA))
  },
  
  #     # small green
  
  "not_covered" = function(x, y, w, h) {
    grid.rect(x, y, w-unit(0.5, "mm"), h-unit(0.5, "mm"),
              gp = gpar(fill = toy_col["not_covered"], col = NA))
  },    # small green
  
  
  
  "arm_level_loss" = function(x, y, w, h) {grid.rect(x, y, w*0.9, h*0.4,
                                                     gp = gpar(fill = toy_col["arm_level_loss"], col = NA))
  },    # small green
  
  
  "arm_level_gain" = function(x, y, w, h) {grid.rect(x, y, w*0.9, h*0.4,
                                                     gp = gpar(fill = toy_col["arm_level_gain"], col = NA))
  },    # small green
  
  
  "wt" = function(x, y, w, h) {grid.rect(x, y, w*0.9, h*0.4,
                                         gp = gpar(fill = toy_col["wt"], col = NA))
  }    # small green
  
)

#make sure toym and toydf are ordered by samples (and in same order as each other)

# call oncoprint function

pdf("Toy Oncoprint.pdf", width = 16, height = 8)

oncoPrint(
  toym,
  show_pct = FALSE,
  alter_fun = toy_alter_fun,
  alter_fun_is_vectorized = FALSE,
  col = toy_col,
  top_annotation = ha_all, 
  row_names_side = "left",
  
  column_split = toydf$Molecular,
  #heatmap_legend_param = heatmap_legend_param,
)
dev.off()
4

0 回答 0