Plotting arrows in ggplot2 for pca-CodePudding

I want to plot arrows in a pca. I can do that with nmds. But for unknown reasons I cannot do it with the pca.

First I tried this:

library(AMR)
ggplot_pca(pca_resources)

The plot is nice but I want the groupings to be color-coded and I do not know how to do that here.

Then I tried it via ggplot2:

ggplot(PCi,aes(x=PC1,y=PC2,color=organ)) 
  geom_point(size=2)   
  geom_abline(intercept=0,slope=0,lty=3,size=0.4)  
  geom_vline(xintercept=0,lty=3,size=0.4)  
  xlab("PC1 (53%)") 
  ylab("PC2 (33%)") 
  theme(#legend.justification=c(1,0),legend.position=c(0.9,0.5),
        panel.grid.major = element_blank(),
        panel.grid.minor = element_blank(),
        axis.text.y=element_text(size=15,family="Arial",color="black"),
        axis.text.x=element_text(size=15,family="Arial",color="black"),
        axis.title.x=element_text(size=15,family="Arial"),
        axis.title.y=element_text(size=15,family="Arial"),
        legend.title=element_text(size=20,family="Arial"),
        legend.text=element_text(size=20,family="Arial"))  
  theme_bw()

Here are the PC scores for the arrows:

structure(c(0.588517849867411, 0.590285718836058, 0.55073730129091, 
0.0435653014731059, -0.267996461049223, -0.199411444560057, 0.433923525535396, 
0.836733736997915, 0.608676467420504, -0.768503848789056, 0.179696493593818, 
-0.081387730002692, -0.459714688976792, -0.14560109997207, 0.690005774717737, 
-0.539772873792367), .Dim = c(4L, 4L), .Dimnames = list(c("FA_18.3.3c_p", 
"FA_18.2.6c_p", "ALA.d13C", "LIN.d13C"), c("PC1", "PC2", "PC3", 
"PC4")))

Here are the data points for the PCA plot:

structure(list(PC1 = c(-1.88530472821989, -2.24714159937733, 
-1.34432344539257, -0.968510753543999, -0.624879049572724, -2.36471016819961, 
-1.16311229527896, -2.61987812001917, -2.08025468016444, -2.68276183422677
), PC2 = c(-1.50284890951331, -0.363086620597548, 1.45264977869589, 
0.721195171897019, 0.93112605562114, -0.906913023282559, -0.283296924307472, 
0.267889285983414, 1.33171541603416, -1.80603301970074), PC3 = c(-0.542337614825214, 
0.646911614334602, 0.0237708210906419, 0.168460682596456, -0.82053283734023, 
-0.101006562583037, -0.151718041641362, 0.574991815166341, 0.564555094889462, 
-0.459509110124471), PC4 = c(0.0280582360577376, 0.863570266385183, 
-0.466390733446598, 1.13230673371715, 0.269061697019256, -0.50980826233187, 
0.181559477299738, -0.12716527942606, 0.107402240826891, -0.785580543274415
), organ = structure(c(4L, 4L, 4L, 4L, 4L, 4L, 4L, 4L, 4L, 4L
), .Label = c("brain", "crustacea", "ephemeroptera", "epilithon", 
"eyes", "fresh.leaves", "liver", "muscle", "plecoptera", "submerged.leaves"
), class = "factor")), row.names = c("3", "4", "5", "6", "7", 
"8", "9", "10", "11", "13"), class = "data.frame")

How can I add the arrows its labels?

Thanks, Nadine

CodePudding user response：

You can reshape the PC co-ordinates and plot them as segments. Here I have used geom_textsegment from the geomtextpath package to easily label the arrows:

library(tidyverse)
library(geomtextpath)

PC_scores %>%
  as.data.frame() %>%
  select(1:2) %>%
  rownames_to_column(var = "var") %>%
  ggplot(aes(0, 0, color = var))  
  geom_hline(yintercept = 0, alpha = 0.2)  
  geom_vline(xintercept = 0, alpha = 0.2)  
  geom_textsegment(aes(xend = PC1, yend = PC2, label = var),
                   arrow = arrow())  
  scale_color_brewer(palette = "Set1")  
  theme_minimal(base_size = 16)  
  theme(legend.position = "none")  
  labs(x = "PC1", y = "PC2")  
  coord_equal()

CodePudding user response：

You could also use the package ggfortify like this:

PCi <- structure(list(PC1 = c(-1.88530472821989, -2.24714159937733, 
                              -1.34432344539257, -0.968510753543999, -0.624879049572724, -2.36471016819961, 
                              -1.16311229527896, -2.61987812001917, -2.08025468016444, -2.68276183422677
), PC2 = c(-1.50284890951331, -0.363086620597548, 1.45264977869589, 
           0.721195171897019, 0.93112605562114, -0.906913023282559, -0.283296924307472, 
           0.267889285983414, 1.33171541603416, -1.80603301970074), PC3 = c(-0.542337614825214, 
                                                                            0.646911614334602, 0.0237708210906419, 0.168460682596456, -0.82053283734023, 
                                                                            -0.101006562583037, -0.151718041641362, 0.574991815166341, 0.564555094889462, 
                                                                            -0.459509110124471), PC4 = c(0.0280582360577376, 0.863570266385183, 
                                                                                                         -0.466390733446598, 1.13230673371715, 0.269061697019256, -0.50980826233187, 
                                                                                                         0.181559477299738, -0.12716527942606, 0.107402240826891, -0.785580543274415
                                                                            ), organ = structure(c(4L, 4L, 4L, 4L, 4L, 4L, 4L, 4L, 4L, 4L
                                                                            ), .Label = c("brain", "crustacea", "ephemeroptera", "epilithon", 
                                                                                          "eyes", "fresh.leaves", "liver", "muscle", "plecoptera", "submerged.leaves"
                                                                            ), class = "factor")), row.names = c("3", "4", "5", "6", "7", 
                                                                                                                 "8", "9", "10", "11", "13"), class = "data.frame")

library(ggfortify)
pca_res <- prcomp(PCi[,-5])
autoplot(pca_res, data = PCi, colour = 'organ', loadings = TRUE, loadings.label = TRUE)

^{Created on 2022-07-30 by the reprex package (v2.0.1)}