更改R中biplot()的轴标签

Question

我正在尝试通过在R中制作一个双图来可视化PCoA{ape}的结果。轴现在得到默认标签轴1和轴2，但我想编辑这个。

这是我尝试过的代码：

biplot(pcoa.ntK, Y=NULL, plot.axes=c(1,2), rn=ntnames,
                           xlabs="PC1 (%)", ylabs="PC2 (%)")

但是标签是不变的。有人能告诉我我哪里做错了吗？我也想编辑一下标题，有什么建议吗？

我的数据：

ntK <- matrix( 
  c(0.00000, 0.01500, 0.01832, 0.02061, 0.01902, 0.01270, 0.02111, 0.01655, 0.01520, 0.01691,
    0.01667, 0.00000, 0.01175, 0.01911, 0.01759, 0.01127, 0.01854, 0.01041, 0.00741, 0.02007,
    0.02432, 0.01404, 0.00000, 0.02551, 0.01972, 0.01838, 0.02505, 0.01484, 0.01391, 0.02687,
    0.01501, 0.01252, 0.01399, 0.00000, 0.01442, 0.01294, 0.01402, 0.01132, 0.01239, 0.01455,
    0.02343, 0.01951, 0.01830, 0.02440, 0.00000, 0.01727, 0.02470, 0.02021, 0.01699, 0.02482,
    0.01320, 0.01054, 0.01439, 0.01847, 0.01457, 0.00000, 0.01818, 0.01366, 0.00977, 0.01394,
    0.02468, 0.01950, 0.02206, 0.02251, 0.02343, 0.02040, 0.00000, 0.02028, 0.01875, 0.02558,
    0.02254, 0.01276, 0.01522, 0.02117, 0.02234, 0.01790, 0.02363, 0.00000, 0.01152, 0.02557,
    0.01804, 0.00792, 0.01244, 0.02019, 0.01637, 0.01116, 0.01904, 0.01004, 0.00000, 0.02099,
    0.01862, 0.01988, 0.02227, 0.02200, 0.02218, 0.01476, 0.02408, 0.02066, 0.01947, 0.00000), 
  nrow=10, 
  ncol=10)

library(ape)
ntnames <- c("A","B","C","D","E","F","G","H","I","J")
pcoa.ntK <- pcoa(ntK)

Answer 1

biplot是一个泛型函数。默认方法和用于使用stats包中的prcomp函数的对象的方法确实允许您指定轴标签和标题，但由于某些原因，编写使用pcoa类的对象调用的方法的人不允许您指定它们。我认为您唯一的选择是编写自己版本的biplot.pcoa (或者请包维护者添加此选项)。

这是对ape包中函数的一个非常快速和肮脏的破解，它可能会做你想做的事情，但不能保证它不会破坏其他东西！

biplot.pcoa <- function (x, Y = NULL, plot.axes = c(1, 2), dir.axis1 = 1, dir.axis2 = 1, 
          rn = NULL, xlabs = NULL, ylabs = NULL, main = NULL, ...) 
{
  k <- ncol(x$vectors)
  if (k < 2) 
    stop("There is a single eigenvalue. No plot can be produced.")
  if (k < plot.axes[1]) 
    stop("Axis", plot.axes[1], "does not exist.")
  if (k < plot.axes[2]) 
    stop("Axis", plot.axes[2], "does not exist.")
  if (!is.null(rn)) 
    rownames(x$vectors) <- rn
  labels = colnames(x$vectors[, plot.axes])
  if (!is.null(xlabs)) labels[1] <- xlabs
  if (!is.null(ylabs)) labels[2] <- ylabs
  diag.dir <- diag(c(dir.axis1, dir.axis2))
  x$vectors[, plot.axes] <- x$vectors[, plot.axes] %*% diag.dir
  if (is.null(Y)) {
    limits <- apply(x$vectors[, plot.axes], 2, range)
    ran.x <- limits[2, 1] - limits[1, 1]
    ran.y <- limits[2, 2] - limits[1, 2]
    xlim <- c((limits[1, 1] - ran.x/10), (limits[2, 1] + 
                                            ran.x/5))
    ylim <- c((limits[1, 2] - ran.y/10), (limits[2, 2] + 
                                            ran.y/10))
    par(mai = c(1, 1, 1, 0.5))
    plot(x$vectors[, plot.axes], xlab = labels[1], ylab = labels[2], 
         xlim = xlim, ylim = ylim, asp = 1)
    text(x$vectors[, plot.axes], labels = rownames(x$vectors), 
         pos = 4, cex = 1, offset = 0.5)
    if (is.null(main)){
      title(main = "PCoA ordination", line = 2.5)
    } else title(main = main, line = 2.5)
  }
  else {
    n <- nrow(Y)
    points.stand <- scale(x$vectors[, plot.axes])
    S <- cov(Y, points.stand)
    U <- S %*% diag((x$values$Eigenvalues[plot.axes]/(n - 
                                                        1))^(-0.5))
    colnames(U) <- colnames(x$vectors[, plot.axes])
    par(mai = c(1, 0.5, 1.4, 0))
    biplot(x$vectors[, plot.axes], U, xlab = labels[1], ylab = labels[2])
    if (is.null(main)) {
    title(main = c("PCoA biplot", "Response variables projected", 
                   "as in PCA with scaling 1"), line = 4)
    } else title(main = main, line = 4)
  }
  invisible()
}

biplot(pcoa.ntK, xlabs = 'My x label', ylabs = 'My y label', main = 'My title')

Answer 2

你可以查看biplot.pcoa的源代码，你会发现修改它并不难。包的作者决定根据输入和绘图的主标题对轴标签进行硬编码。下面是一个修改后的版本，它将首先检查是否使用了xlab、ylab和main的值，然后再使用预定义的值：

biplot.pcoa <- function (x, Y = NULL, plot.axes = c(1, 2), dir.axis1 = 1, dir.axis2 = 1, 
          rn = NULL, ...) 
{
  k <- ncol(x$vectors)
  if (k < 2) 
    stop("There is a single eigenvalue. No plot can be produced.")
  if (k < plot.axes[1]) 
    stop("Axis", plot.axes[1], "does not exist.")
  if (k < plot.axes[2]) 
    stop("Axis", plot.axes[2], "does not exist.")
  if (!is.null(rn)) 
    rownames(x$vectors) <- rn
  args <- list(...)
  labels = ifelse(c("xlab", "ylab") %in% names(args), c(args$xlab, args$ylab), colnames(x$vectors[, plot.axes]))
  diag.dir <- diag(c(dir.axis1, dir.axis2))
  x$vectors[, plot.axes] <- x$vectors[, plot.axes] %*% diag.dir
  if (is.null(Y)) {
    limits <- apply(x$vectors[, plot.axes], 2, range)
    ran.x <- limits[2, 1] - limits[1, 1]
    ran.y <- limits[2, 2] - limits[1, 2]
    xlim <- c((limits[1, 1] - ran.x/10), (limits[2, 1] + 
                                            ran.x/5))
    ylim <- c((limits[1, 2] - ran.y/10), (limits[2, 2] + 
                                            ran.y/10))
    par(mai = c(1, 1, 1, 0.5))
    title <- ifelse("main" %in% names(args), args$main, "PCoA ordination")
    plot(x$vectors[, plot.axes], xlab = labels[1], ylab = labels[2], 
         xlim = xlim, ylim = ylim, asp = 1, 
         main = title)
    text(x$vectors[, plot.axes], labels = rownames(x$vectors), 
         pos = 4, cex = 1, offset = 0.5)
    #title(main = "PCoA ordination", line = 2.5)
  }
  else {
    n <- nrow(Y)
    points.stand <- scale(x$vectors[, plot.axes])
    S <- cov(Y, points.stand)
    U <- S %*% diag((x$values$Eigenvalues[plot.axes]/(n - 
                                                        1))^(-0.5))
    colnames(U) <- colnames(x$vectors[, plot.axes])
    par(mai = c(1, 0.5, 1.4, 0))
    title <- ifelse("main" %in% names(args), args$main, c("PCoA biplot", "Response variables projected", 
                                                              "as in PCA with scaling 1"))
    biplot(x$vectors[, plot.axes], U, xlab = labels[1], ylab = labels[2], main = title)
    # title(main = c("PCoA biplot", "Response variables projected", 
    #                "as in PCA with scaling 1"), line = 4)
  }
  invisible()
}

然后：

biplot(pcoa.ntK, Y=NULL, plot.axes=c(1,2), rn=ntnames,
   xlab="PC1 (%)", main = "Main Title")

​

​

记住这不会改变原来的功能，所以你需要在每次加载包的时候加载这个修改过的版本，并且需要像这样设置标签。