Distance and Parsimony methods (Part 3: computer lab)

Software R package ape: Distance-based methods

ape is one of the most widely used phylogenetic software
It is an R package and it has a huge variety of functions
In particular, today we will use it for distance-based tree estimation methods
Full documentation
We will follow this great tutorial

In-class group dynamic

Time: 20 minutes

Instructions:

Follow the R commands to obtain a ME tree from the sample data (or your own data!). The commands are listed in the PDF tutorial that we are using as guideline, in our reproducible script notebook-log.md and on the following slides.
After the allotted time, we will compare our work all together.

Solution

1) Installing necessary packages:

install.packages("adegenet", dep=TRUE)
install.packages("phangorn", dep=TRUE)

2) Loading the packages

library(ape)
library(adegenet)
library(phangorn)

3) Loading the sample data

dna <- fasta2DNAbin(file="http://adegenet.r-forge.r-project.org/files/usflu.fasta")

4) Computing the genetic distances. They choose a Tamura and Nei 1993 model which allows for different rates of transitions and transversions, heterogeneous base frequencies, and between-site variation of the substitution rate (more on Models of Evolution).

D <- dist.dna(dna, model="TN93")

5) Get the NJ tree

tre <- nj(D)

6) Before plotting, we can use the ladderize function which reorganizes the internal structure of the tree to get the ladderized effect when plotted

tre <- ladderize(tre)

7) We can plot the tree

plot(tre, cex=.6)
title("A simple NJ tree")

How do we interpret the tree?

Summary of Software R package ape

Main distance functions:

nj (ape package): the classical Neighbor-Joining algorithm.
bionj (ape): an improved version of Neighbor-Joining: Gascuel 1997. It uses information on variances of evolutionary distances
fastme.bal and fastme.ols (ape): minimum evolution algorithms: Desper and Gascuel, 2002
hclust (stats): classical hierarchical clustering algorithms including single linkage, complete linkage, UPGMA, and others.

Software: R package phangorn: Parsimony-based methods

phangorn is another widely used phylogenetic software
It is an R package and it has a huge variety of functions
In particular, today we will use it for parsimony-based tree estimation methods
Full documentation
We will follow this great tutorial

In-class group dynamic

Time: 20 minutes

Instructions:

Follow the R commands to obtain a MP tree from the sample data (or your own data!). The commands are listed in the PDF tutorial that we are using as guideline or in our reproducible script notebook-log.md or on the following slides.
After the allotted time, we will compare our work all together.

Solution

1) Installing necessary packages (if you have not installed them for the distance section above)

install.packages("adegenet", dep=TRUE)
install.packages("phangorn", dep=TRUE)

2) Loading

library(ape)
library(adegenet)
library(phangorn)

3) Loading the sample data and convert to phangorn object:

dna <- fasta2DNAbin(file="http://adegenet.r-forge.r-project.org/files/usflu.fasta")
dna2 <- as.phyDat(dna)

4) We need a starting tree for the search on tree space and compute the parsimony score of this tree (422)

tre.ini <- nj(dist.dna(dna,model="raw"))
parsimony(tre.ini, dna2)

5) Search for the tree with maximum parsimony:

> tre.pars <- optim.parsimony(tre.ini, dna2)
Final p-score 420 after  2 nni operations

6) Plot tree:

plot(tre.pars, cex=0.6)

Further learning:

Continue the parsimony steps in the PDF tutorial on the same sample data.

Homework

See the details of the Distance and Parsimony HW in here.