R如何在heatmap.2中的列上方构建倾斜的列标题:将(文本)图传递到布局? [英] R How to build angled column headings above columns in heatmap.2: pass (text) plot to the layout?

问题描述

我与想要的热图非常接近，但是我一直在努力寻找标题问题.我希望每列顶部都有倾斜的标题(例如45度或50度).我已经抑制了树状图(对于列和行)，并使用矩阵的第一列(深度)作为行的标签.

I am very close to the heatmap I want, but I have been struggling for several days to figure out the headings problem. I want angled headings (such as 45 or 50 degrees) at the top of each column. I have already suppressed the dendrograms (for both columns and rows), and used the first column of my matrix (depth) as labels for the rows.

从许多关于此问题的搜索中，我发现mtext不能帮上忙，因为文本无法在mtext中旋转.然后我以为我可以用文本来做，但是列标签被覆盖到了热图本身上.当单词到达热图布局空间的边缘时，它们消失"(被遮盖).因此，我检查了热图使用的布局(非常感谢@Ian Sudbery)，在我看来，我真正需要的是在布局中用于列标题的专用空间.我可以使用layout函数分配该空间，我已经在下面的代码中完成了分配.但是我认为下一步可能涉及进入heatmap.2代码. Heatmap.2调用了四个图(我抑制了两个图，树状图).我如何称其为第五情节?如果可能的话，我怎么告诉我要使用文本作为我的第五个情节"，其中文本是我的列标题，旋转了50度?

From my many, many searches on this issue, I see that mtext won't help me because the text cannot be rotated in mtext. Then I thought I could do it with text, but the column labels get overwritten onto the heatmap itself; they "disappear" (get covered) when the words reach the edge of the heatmap layout space. So I examined the layout used by heatmap (thanks very much to @Ian Sudbery), and it occurred to me that what I really need is a dedicated space in the layout for my column headings. I can allocate that space using the layout function, and I have done so in the code below. But I think the next step may involve getting inside the heatmap.2 code. Heatmap.2 calls four plots (two of which I have suppressed, the dendrograms). How do I make it call a fifth plot? And if that is possible, how do I tell it that I want to use text as my fifth "plot", where the text is my column headings, rotated 50deg?

非常感谢您的任何想法.请原谅我在下面的代码中提供示例数据的笨拙方式；我是R的新手，通常不知道做大多数事情的最优雅的方式.

Thanks very much for any ideas. Please forgive the clumsy way I've provided sample data in the code below; I am new to R and generally do not know the most elegant way to do most things.

par(oma=c(0.5,0.5,.5,0.5))  # outer margins of whole figure

lmat = rbind(c(3,5,5),c(2,1,1),c(0,4,0))
lwid = c(.08,.9, .1)
lhei = c(1,4,1.5)

Depth<-c("0m","20m","40m","60m","80m","100m")
Sept2008<-c(3,6,8,10,15,16)
March2010<-c(10,12,11,13,12,11)
Sept2010<-c(5,6,NA,8,11,13)
March2011<-c(4,6,10,NA,14,14)
Sept2011<-c(2,5,3,9,16,12)

heatmap_frame=data.frame(Depth=Depth,Sept2008=Sept2008,March2010=March2010,Sept2010=Sept2010, March2011=March2011, Sept2011=Sept2011)
row.names(heatmap_frame)<-heatmap_frame$Depth
heatmap_frame<-heatmap_frame[,-1]
heatmap_matrix <- as.matrix(heatmap_frame)

labCol=c("Sept 2008","March 2010","Sept 2010","March 2011","Sept 2011")
cexCol=1.1

heatmap <- heatmap.2(heatmap_matrix, dendrogram="none", trace="none",Rowv=NA, Colv=NA, 
                 col = brewer.pal(9,"Blues"), scale="none", margins=c(2,5),labCol="",
                 lmat=lmat, lwid=lwid,lhei=lhei, density.info="none",key=TRUE)

# want to plot a fifth area, to use for col labels
# don't know how to pass a text line to the heatmap.2/layout/matrix to print as my fifth plot

mtext("Use for main title", side=3,outer=F,line=2.75, font=2, cex=1.25)

# testing the text function; did not work as desired
#text(x=1, y=1, labels="Label_1",xpd=T)
text(x=c(0,.2,.4,.6,.8), y=0.95, pos=3, srt=50, labels=labCol,xpd=T, cex=1)

推荐答案

这是一种黑客程序，它不涉及拆开heatmap.2复杂的代码:

Here's a hack that doesn't involve pulling apart the convoluted code of heatmap.2:

  pos2 <- locator() #will return plotting coordinates after doing this:
  # Shift focus to the graphics window by clicking on an edge 
  # Left-Click once where you want the first label to be centered
  # Left-click again on the point where you want the last label centered
  # Right-Click, then return focus to the console session window

 pos2 <- structure(list(x = c(0.27149971320082, 0.858971646016485), 
                        y = c(0.861365598392473, 0.857450478257082)),
                       .Names = c("x", "y"))

  text(x=seq(pos2$x[1], pos2$x[2], len=5), y=rep(pos2$y[1],5)  ,
       srt=50, xpd=TRUE, adj = 0,
       labels=c("Sept 2008","March 2010","Sept 2010",
                "March 2011","Sept 2011") )

我不知道您是否真的需要xpd，因为看来heatmap.2完成后它会将窗口返回其原始坐标:[0,1] x [0,1]

I don't know if you actually need the xpd in there, since it appears that after heatmap.2 is finished it returns the window to its native coordinates: [0,1]x[0,1]

这篇关于R如何在heatmap.2中的列上方构建倾斜的列标题:将(文本)图传递到布局?的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！