r - 如何从 R pheatmap 获取色标值(行 z 分值)

Question

我有以下数据框:

library(pheatmap)
library(RColorBrewer)

dat <- structure(c(36.28516213, -9.184523335, 5.716034994, 32.82180488, 
-18.5467127, 22.79930301, -24.79075287, -49.23768715), .Dim = c(2L, 
4L), .Dimnames = list(c("motif_a", "motif_b"), c("cow", "dog", 
"snake", "cat")))

dat
#>               cow       dog     snake       cat
#> motif_a 36.285162  5.716035 -18.54671 -24.79075
#> motif_b -9.184523 32.821805  22.79930 -49.23769

并使用以下热图代码:

p <- pheatmap(
  dat,
  color = colorRampPalette(rev(RColorBrewer::brewer.pal(n = 7, name = "RdYlBu")))(1000),
  clustering_method = "ward.D2",
  border_color = "grey60",
  clustering_distance_columns = "correlation",
  clustering_distance_rows = "correlation",
  fontsize_row = 6,
  cluster_rows = TRUE,
  cluster_cols = TRUE,
  scale = "row"
)

我可以得到这个情节。

正如图中所指出的，我怎样才能得到相应的值热图中的每个点都对应于右侧色标。

例如，cow motif_b 的值为 -9.184523，但右手色标给出的值介于 0 到 -0.5 之间。我怎样才能得到这些值？

最终结果是一个数据框，其中包含已转换的值。我该怎么做？

Answer 1

当您指定 clustering_distance_xxx = "correlation" 时，您要求 pheatmap 绘制介于 -1 和 1 之间的 Pearson 相关性。运行 ?pheatmap 了解更多详情。

如果您从代码中删除这些部分，那么您将得到您期望的结果

library(pheatmap)
library(RColorBrewer)

dat <- structure(c(36.28516213, -9.184523335, 5.716034994, 32.82180488, 
    -18.5467127, 22.79930301, -24.79075287, -49.23768715), .Dim = c(2L, 
    4L), .Dimnames = list(c("motif_a", "motif_b"), c("cow", "dog", 
    "snake", "cat")))

myCol <- colorRampPalette(rev(RColorBrewer::brewer.pal(n = 7, name = "RdYlBu")))(8)

p <- pheatmap(
  dat,
  color = myCol,
  clustering_method = "ward.D2",
  border_color = "grey60",
  clustering_distance_columns = "correlation",
  clustering_distance_rows = "correlation",
  fontsize_row = 6,
  cluster_rows = TRUE,
  cluster_cols = TRUE,
  scale = "row"
)

p1 <- pheatmap(
  dat,
  color = myCol,
  clustering_method = "ward.D2",
  border_color = "grey60"
)

由 reprex package 创建于 2018-03-27 (v0.2.0).