Codes used for getting sequences for phylogenetic tree and making Tanglegrams

1. Getting sequences from csv file

a. installing and loading packages

install.packages("seqinr")
install.packages("ape")
install.packages("tidyverse")
install.packages("readxl")
library(seqinr)
library(ape)
library(dplyr)
library(tidyverse)
install.packages("xlsx")
library(xlsx)

b. reading sequences from file

setwd(dir ="/Users/amrit/Desktop/order_name")

order_name_matk_num <-read.csv("order_name_matk.csv")
order_name_matk <- read.GenBank(order_name_matk_num$matk)

order_name_rbcL_num <- read.csv("order_name_rbcl.csv")
order_name_rbcL <- read.GenBank(order_name_rbcL_num$rbcL)

order_name_trnh_num <- read.csv("order_name_trnh.csv")
order_name_trnh <- read.GenBank(order_name_trnh_num$trnh_psba)

its1_num <- read.csv("order_name_its1.csv")
its1 <- read.GenBank(its1_num$its)

its2_num <- read.csv("order_name_its2.csv")
its2 <- read.GenBank(its2_num$its2)

c. adding accession names to species names

matk_IDs <- paste(attr(order_name_matk, "species"), names(order_name_matk), sep = "_matK_")
rbcL_IDs <- paste(attr(order_name_rbcL, "species"), names(order_name_rbcL), sep = "_rbcL_")
trnh_IDs <- paste(attr(order_name_trnh, "species"), names(order_name_trnh), sep = "_trnh_")
its1_IDs <- paste(attr(its1, "species"), names(its1), sep = "_its1_")
its2_IDs <- paste(attr(its2, "species"), names(its2), sep = "_its2_")

matk_IDs
rbcL_IDs
trnh_IDs
its1_IDs
its2_IDs

d. writing them in fasta format

write.dna(order_name_matk, file = "order_name_matk.fasta", format = "fasta", append = FALSE)
write.dna(order_name_rbcL, file = "order_name_rbcL.fasta", format = "fasta", append = FALSE)
write.dna(order_name_trnh, file = "order_name_trnh.fasta", format = "fasta", append = FALSE)
write.dna(its1, file = "its1.fasta", format = "fasta", append = FALSE)
write.dna(its2, file = "its2.fasta", format = "fasta", append = FALSE)

e. rewriting with more information

order_name_matk_seqinr <- read.fasta(file = "order_name_matk.fasta", seqtype = "DNA", as.string = TRUE, forceDNAtolower = FALSE)
order_name_rbcL_seqinr <- read.fasta(file = "order_name_rbcL.fasta", seqtype = "DNA", as.string = TRUE, forceDNAtolower = FALSE)
order_name_trnh_seqinr <- read.fasta(file = "order_name_trnh.fasta", seqtype = "DNA", as.string = TRUE, forceDNAtolower = FALSE)
its1_seqinr <- read.fasta(file = "its1.fasta", seqtype = "DNA", as.string = TRUE, forceDNAtolower = FALSE)
its2_seqinr <- read.fasta(file = "its2.fasta", seqtype = "DNA", as.string = TRUE, forceDNAtolower = FALSE)

f. wrting fasta files

write.fasta(sequences = order_name_matk_seqinr, names = order_name_matk_num$speciesnames, file.out = "order_name_matk_sequences.fasta")
write.fasta(sequences = order_name_rbcL_seqinr, names = order_name_rbcL_num$speciesnames, file.out = "order_name_rbcL_sequences.fasta")
write.fasta(sequences = order_name_trnh_seqinr, names = order_name_trnh_num$speciesnames, file.out = "order_name_trnh_sequences.fasta")
write.fasta(sequences = its1_seqinr, names = its1_num$speciesnames, file.out = "its1_sequences.fasta")
write.fasta(sequences = its2_seqinr, names = its2_num$speciesnames, file.out = "its2_sequences.fasta")

2. Creating cophyloplots for each order

# use the previously installed packages
library(ape)
library(phytools)

a. reading and preparing the trees

order_name_chloroplast_tree <- ape::read.tree("~/Desktop/Manuscript/order_name/order_name.tre")
order_name_nuclear_tree <- ape::read.tree("~/Desktop/Manuscript/order_name/order_name_its2.tre") 

order_name_chloroplast_tree <- compute.brlen(order_name_chloroplast_tree)
order_name_nuclear_tree <- compute.brlen(order_name_nuclear_tree)

order_name_chloroplast_tree <- midpoint_root(order_name_chloroplast_tree)
order_name_nuclear_tree <- midpoint.root(order_name_nuclear_tree)

b. defining association matrix

order_name_association <- cbind(order_name_chloroplast_tree$tip.label, order_name_chloroplast_tree$tip.label)

c. creating cophylo object

obj <- cophylo(order_name_chloroplast_tree, order_name_nuclear_tree, assoc = order_name_association, print = TRUE)

d. plot to see how it looks

plot(order_name_chloroplast_tree, lwd = 2, pts = 1, link.type = "curved", link.lwd = 2, link.lty = "solid", link.col = make.transparent("blue", .25), fsize = 1, rotate.multi = TRUE)
plot(order_name_nuclear_tree, lwd = 2, pts = 1, link.type = "curved", link.lwd = 2, link.lty = "solid", link.col = make.transparent("blue", .25), fsize = 1, rotate.multi = TRUE)

e. save to PDF

pdf("~/Desktop/Manuscript/order_name/order_name_cophylo_plot.pdf", width = 10, height = 10)
plot(obj,
     lwd = 2, pts = 1,
     link.type = "curved",
     link.lwd = 2,
     link.lty = "solid",
     link.col = make.transparent("blue", 0.25),
     fsize = 1,
     rotate.multi = TRUE)
dev.off()