Chapter 8 Building Trees
8.1 Background on tree building
Watch the following videos on phylogenetic tree construction
- Video 1: review
- Video 2: Tree building (UPGMA)
- Video 6: Tree building (NJ)
- Video 5: Tree building (parsimony)
Don’t worry about the details - I just want you to get a sense of different methods of tree building rather than viewing this process as a “black box” where you just put your data into a computer program and get an answer. For a summary read A brief guide to the different methods of phylogenetic tree construction
8.2 Building trees from your data
8.2.1 Likelihood
We’ll take two different approaches to build phylogenies from your data. First, we’ll use software that calculates the likelihood that the data fit the phylogeny. The software will try lots of different phylogenies and pick the case where the data are most likely. When the software runs it will also try resampling the data to see how much support we have for each relationship we estimate.
- Use
nano
to openrunRAxML.sh
in theslurm_example_scripts
folder. - Change all lines marked CHANGE
- The path to the alignment file should be the relative path to one of the files labeled alignment_pi_m1_nogap.phylip-relaxed (yours may have a different number after m)
- Submit the script as a slurm job.
8.3 Visualizing trees
- Check that a file starting with
RAxML_bipartitions
is in your slurm folder by listing files (i.e. in the same location from which you submitted your scripts). - Use
nano
to openviewTree.slurm
. - Change all lines marked CHANGE
- Follow the instructions for transferring files from Unity to obtain the pdf output for your tree (it should have the same name as the input tree file followed by pdf)
8.4 What does my tree mean?
Interpret your tree using what you know about the group and how to read trees.
8.5 Other tree building methods
Another method for building phylogenies relies on selecting four species, and determining the relationships of these species based on examining different patterns of sites. Four species are much easier to work with than more as there are only three possible ways these species can be related. We can look at how often the grouping of sites matches each hypotheized why species are related. You can read the full explanation (with visuals) but don’t worry if it’s complicated. This method is implemented in SVDQuartets.
- Change the necessary files and information in
runSVDQ.sh
and submit it. - Visualize your tree using the same process as for the RAxML tree but noting that you are looking for a file ending in .tre in the folder containing your input alignment.
- Download and examine your results
8.6 Alternative methods of support
The bootstrap approach automatically implemented by RAxML is one method of determining support. However, bootstrap resampling can have limitations when using larger datasets. An alternative approach is to examine the proportion of sites that support each alternative hypothesis given a tree, similar to the SVDQ approach.
- Calculate SCF support by editing the
runSCF.sh
script and inputting your tree from either RAxML or SVDQ and your alignment. - If you estimated different trees using the above methods you may try more than one.
- Visualize and download as for other trees.