Annotate in ggplot2 does not honor newline is a pa

2019-05-23 07:55发布

问题:

Question

How do I get paste and parse in annotate of ggplot2 to honor a newline (\n) character?

Problem and MWE

I am trying to reproduce in ggplot2 a stressplot of an NMDS analysis using metaMDS in package vegan. Here is my MWE, followed by the resulting graph.

library(ggplot2)
library(tibble)
library(vegan)

set.seed(42) # for reproducibility

data(dune)
fit <- metaMDS(dune)
tib <- tibble(fit$diss, fit$dist, fit$dhat)
colnames(tib) <- c("diss", "dist", "dhat")
stress <- fit$stress
coord_x <- min(tib$diss)
coord_y <- max(tib$dist)
nonmetric_r2 <- round(1 - stress * stress, digits = 3)
linear_r2 <- round(summary(lm(fit$dist~fit$dhat))$adj.r.squared, 3)

## How do I get the newline character to be honored?
nonmetric_label = paste0("Non-metric~fit~italic(R)^2 ==", nonmetric_r2, "~\n Linear~fit~italic(R)^2 ==", linear_r2)

ggplot(tib,
       aes(x = diss, y = dist)) +
  geom_point(color = "blue") +
  geom_line(aes(x = diss, y = dhat), color = "red") +
  annotate(
    geom = "text",
    x = coord_x,
    y = coord_y,
    hjust = 0,
    #vjust = 1,
    label = nonmetric_label, parse = TRUE) +
  labs(x = "Observed Dissimilarity",
       y = "Ordination Distance")

The single annotated line above should be on two separate lines, as shown below (from stressplot(fit)).

The offending line is

nonmetric_label = paste0("Non-metric~fit~italic(R)^2 ==", nonmetric_r2, "~\n Linear~fit~italic(R)^2 ==", linear_r2)

If I do not include the tilde before \n, then everything after the newline character disappears. I have tried various combinations of tilde placement and placing '\n~Linear~fit' in additional single quotes and back-ticks.

How do I get the desired annotation to appear on two lines?

回答1:

One approach would be to use a vector of strings as a label and a vector of coordinates which will mach the desired annotation:

nonmetric_label = c(paste0("Non-metric~fit~italic(R)^2 ==", nonmetric_r2),
                    paste0("Linear~fit~italic(R)^2 ==", linear_r2)) 

ggplot(tib,
       aes(x = diss, y = dist)) +
  geom_point(color = "blue") +
  geom_step(aes(x = diss, y = dhat), color = "red", direction = "vh") +
  annotate(
    geom = "text",
    x = coord_x,
    y = c(coord_y, 0.95*coord_y),
    hjust = 0,
    #vjust = 1,
    label = nonmetric_label, parse = TRUE) +
  labs(x = "Observed Dissimilarity",
       y = "Ordination Distance")

As per suggestion of Jari Oksanen I have changed geom_line to geom_step. To match the output of stressplot(fit) an additional argument direction = "vh" is needed.



标签: r ggplot2 vegan