Tanglegram

This guide is about making co-phylogeny tanglegrams.

Instruction

---

Part 1: Manual tanglegram with ggtree

Step1: Preparation

1st step - install or call libraries.

Input

if (!require("pacman")) install.packages("pacman")

pacman::p_load(ggplot2, ggtree, phangorn, dplyr)

2nd step - set the working directory.

Input

main_dir <- dirname(rstudioapi::getSourceEditorContext()$path)
setwd(main_dir)

---

Step 2: Generate the trees and prepare them

Create a directory for images.

Input

dir.create("imgs", showWarnings = TRUE, recursive = FALSE, mode = "0777")

Generate 2 small random phylogenetic trees.

Input

# Generate two random trees with the same set of tips (for demonstration)
set.seed(34)
tips <- LETTERS[1:10]  # Shared tip labels

tree1 <- rtree(n = 10, tip.label = tips)  # Random tree 1
tree2 <- rtree(n = 10, tip.label = tips)  # Random tree 2

# Ensure that both trees have the same tip labels for co-phylogeny plotting
tree2$tip.label <- tree1$tip.label  # Sync labels

Display them.

Input

# Plot tree1 (left) and tree2 (right)
t1 <- ggtree(tree1) + 
  geom_tiplab(offset = .1)

t2 <- ggtree(tree2) + 
  geom_tiplab(offset = .1)

# Combine the two plots side by side
two_trees <- t1 + t2
two_trees

Output

These are the trees we will be working with.
Now let's save them.

Input

ggsave("imgs/two_trees.png", two_trees, width = 10, height = 8, dpi = 600)

---

Step 3: Rotate the 2nd tree

Of course we can rotate the 2nd tree simply using scale_x_reverse().

Input

# Rotate the 2nd tree
t2_rotated <- ggtree(tree2) + 
  geom_tiplab(offset = -.1) +
  scale_x_reverse()

# Combine the two plots side by side
two_trees_w_ro <- t1 + t2_rotated
two_trees_w_ro

Output

Let's save these trees.

Input

ggsave("imgs/two_trees_w_ro.png", two_trees_w_ro, width = 10, height = 8, dpi = 600)

Looking good! But to manually create the tanglegram we need to adjust the coordinates of labels. And if we use scale_x_reverse() it will be impossible to make the tanglegram. That is why we need to go by more difficult and not obvious way - by manually horizontally mirroring the tree...

Now we are going to grab the backend data frame from both trees and update the tree 2 data frame x-coordinate.

Input

data_tree_1 <- t1$data
data_tree_2 <- t2$data

data_tree_1$tree <-'t1'
data_tree_2$tree <-'t2'

And now we will use the max(d2$x) - d2$x + max(d1$x) + max(d1$x)*0.3 equation to update x coordinates for the 2nd tree. This equation is not universal, 0.3 is not constant, you can fiddle with different values depending on the branch length unit of your tree to get good visualization.

Input

data_tree_2$x <- max(data_tree_2$x) - data_tree_2$x + max(data_tree_1$x) +  max(data_tree_1$x)*0.3

Now we display both trees with manually rotated 2nd tree.

Input

two_trees_w_ro_2 <- t1 +
  geom_tree(data=data_tree_2) +
  geom_tiplab(data = data_tree_2, offset = - 0.2)
two_trees_w_ro_2

Output

Looking same good! And let's save these twoo trees.

Input

ggsave("imgs/two_trees_w_ro_2.png", two_trees_w_ro_2, width = 10, height = 8, dpi = 600)

---

Step 4: Plotting the tanglegram

Now let’s merge data_tree_1 and data_tree_2 to data_combined dataframe so that we can use the coordinates of the tips for making connections between both of the trees.

Input

data_combined <- bind_rows(data_tree_1, data_tree_2) %>% 
  filter(isTip == TRUE)

Now, we conditionally join the tips of both trees. Connected tips will represent the same isolates.

Input

ggtree_tanglegram_1 <- two_trees_w_ro_2 +
  geom_line(aes(x, y, group=label), data=data_combined, color='#009E73')

ggtree_tanglegram_1

Output

Hooray! Let's save this tanglegram. But... the connecting lines overlap the tip labels. Now we must solve this problem!

Input

ggsave("imgs/ggtree_tanglegram_1.png", ggtree_tanglegram_1, width = 10, height = 8, dpi = 600)

Copy the data_combined dataset with coordinates and increase x coordinates for the 1st tree by 0.3 and decrease x coordinated for the 2nd tree by 0.3.

Input

data_combined_padding <- data_combined

data_combined_padding$x[data_combined_padding$tree == 't1'] <- data_combined_padding$x[data_combined_padding$tree == 't1'] + 0.3
data_combined_padding$x[data_combined_padding$tree == 't2'] <- data_combined_padding$x[data_combined_padding$tree == 't2'] - 0.3

Finally, let's make one last visualization.

Input

ggtree_tanglegram_2 <- two_trees_w_ro_2 +
  geom_line(aes(x, y, group=label), data=data_combined_padding, color='#009E73')

ggtree_tanglegram_2

Output

Just perfect! And let's save it.

Input

ggsave("imgs/ggtree_tanglegram_2.png", ggtree_tanglegram_2, width = 10, height = 8, dpi = 600)

---

Part 2: Tanglegram with ape and cophyloplot()

Step 1: Preparation

1st and the last step - install or call libraries.

This time the preparation is simple - just install/call ape library.

Input

pacman::p_load(ape)

---

Step 2: Small tanglegram

Generate 2 small random phylogenetic trees.

Input

TreeA_S <- rtree(10)
TreeB_S <- rtree(10)

Now let's take a look at these trees.

Input

# Plot tree1 (left) and tree2 (right)
t1s <- ggtree(TreeA_S) + 
  geom_tiplab(offset = .1)

t2s <- ggtree(TreeB_S) + 
  geom_tiplab(offset = .1)

# Combine the two plots side by side
two_trees_s <- t1s + t2s
two_trees_s

Output

Interesting trees. Let's save them.

Input

ggsave("imgs/two_trees_s.png", two_trees_s, width = 10, height = 8, dpi = 600)

Create the association matrix.

Input

association <- cbind(TreeB_S$tip.label, TreeB_S$tip.label)

Now let's display the tanglegram.

Input

cophyloplot(TreeA_S, TreeB_S, assoc = association, length.line = 4, space = 28, gap = 3)

Output

Wow! Let's save this plot.

Input

png("imgs/ape_small_tanglegram.png", res = 600)
cophyloplot(TreeA_S, TreeB_S, assoc = association, length.line = 4, space = 28, gap = 3)
dev.off()

---

Step 3: Large tanglegram

cophyloplot() function of ape package workid just fine for the small trees. But will it work for large trees! Let's find it out!

Generate 2 large random phylogenetic trees.

Input

TreeA_L <- rtree(100)
TreeB_L <- rtree(100)

Now let's take a look at these trees.

Input

# Plot tree1 (left) and tree2 (right)
t1l <- ggtree(TreeA_L) + 
  geom_tiplab(size = 1.5, offset = .1)

t2l <- ggtree(TreeB_L) + 
  geom_tiplab(size = 1.5, offset = .1)

# Combine the two plots side by side
two_trees_l <- t1l + t2l
two_trees_l

Output

Wow! Suppose cophyloplot() will not do good with such a large trees. Anyway let's save them.

Input

ggsave("imgs/two_trees_l.png", two_trees_l, width = 10, height = 8, dpi = 600)

Create the association matrix.

Input

association <- cbind(TreeB_L$tip.label, TreeB_L$tip.label)

Now let's display the tanglegram.

Input

cophyloplot(TreeA_L, TreeB_L, assoc = association, length.line = 4, space = 28, gap = 3)

Output

That looks awful... Anyway let's save it...

Input

png("imgs/ape_large_tanglegram.png", res = 600)
cophyloplot(TreeA_L, TreeB_L, assoc = association, length.line = 4, space = 28, gap = 3)
dev.off()

---

Part 3: Tanglegram with dendextend

For more guides dendextend package please visit official guide

How do we make large tanglegram look good? Use dendextend!

Step 1: Preparation

1st and the last step - install or call libraries.

This time the preparation is also simple - just install/call dendextend library.

Input

pacman::p_load(dendextend)

Set another seed to make the trees look different.

Input

set.seed(42)

---

Step 2: Creating dendrograms

tanglegram function in dendextend package uses ultrametric trees. That's why to generate 2 large random phylogenetic trees we will use rcoal() function.

Input

TreeA_L_2 <- rcoal(100)
TreeB_L_2 <- rcoal(100)

Now let's take a look at these trees.

Input

# Plot tree1 (left) and tree2 (right)
t1l2 <- ggtree(TreeA_L_2) + 
  geom_tiplab(size = 1.5, offset = .1)

t2l2 <- ggtree(TreeB_L_2) + 
  geom_tiplab(size = 1.5, offset = .1)

# Combine the two plots side by side
two_trees_l_2 <- t1l2 + t2l2
two_trees_l_2

Output

Yeah. Looking massive. Save it.

Input

ggsave("imgs/two_trees_l_2.png", two_trees_l_2, width = 10, height = 8, dpi = 600)

---

Step 3: Plotting tanglegram

Now let's use tanglegram function!

Input

tanglegram(TreeA_L_2, TreeB_L_2,
           sort = TRUE,
           common_subtrees_color_lines = FALSE,
           highlight_distinct_edges  = FALSE,
           highlight_branches_lwd = FALSE)

Output

Finally! That large tanglegram looks great! Let's save it.

Input

png("imgs/dendextend_large_tanglegram.png", width = 9, height = 16, res = 600)
tanglegram(TreeA_L_2, TreeB_L_2,
           sort = TRUE,
           common_subtrees_color_lines = FALSE,
           highlight_distinct_edges  = FALSE,
           highlight_branches_lwd = FALSE)
dev.off()