Generating multiple sequence alignments and trees with Clustal, Phylip and PAUP

 

You are being provided with three data sets for analysis.  Two of the data sets are protein and one is nucleotide. This assignment has 5 parts.

 

1.     Making multiple sequence alignments  (MSAs) for your data sets (or you can use alignments created with Urmilla).

2.     Studying your alignments to see what you can learn from them

3.     Use MSAs for phylogenetic analysis with Phylip

4.     Use MSAs for bootstrap analysis

5.     Interpret results.

 

Before you begin, prepare your work space:

Make a new directory for this assignment "mkdir phylo"  Change into this new directory and  make three additional directories. If you are using alignments created with Urmilla, make directories for these files. One for each MSA. Label them with the data type, e.g. nucleotide or protein

 

cd phylo

mkdir cs-prot

mkdir actin-prot

mkdir actin-nuc


             

Now, "cd" into each of these directories and  copy over the original text files used to make the MSA with clustalX and the resulting phylip formatted alignment file from wherever you put the files from Urmilla's ecercise. Or, download the files provided to you on the web. You will find thm in the data directory.  Be sure to put the files in the correct directories that you created for them.

 

cd cs-prot

 download cs.txt

 download cs.phy


cd ../actin-prot

 download actinpep.txt       NOTE: there is no alignment you will need to create it


cd ../actin-nuc

 download actinnuc.txt       NOTE: there is no alignment you will need to creat it



 ClustalX takes a text file of fasta formatted sequences and creates an interleaved multiple sequence alignment out of them. You have control over the output sequence formats. By default the program will create ".aln" and ".dnd" formats. You will need to add the ".phy"  formats.  Do this within the correct directories. This should take only 4-5 minutes each.


Launch clustalX by typing "clustalx" on the command line .

 

When you have finished the alignments the resulting MSAs are already stored in the files that were created.   View your alignment before quitting.

 

2) Study your alignments.  What do you see?  Please answer the following questions about each of your alignments (except question ÒeÓ which only applied to one data set).


The actin data sets contain the same sequences, as either DNA or protein.  Yeast is your outgroup. The drosophila sequences represent the protostome organisms and the all of the orther organisms are deuterostomes, with tunicates (halocynthia) and Echinoderms (strongylocentrotus purpuratus) as representatives of the early branches. You are now ready to begin.  Please perform all analyses related to a particular data set within the directory for that data set.



1.     Are there lots of gaps?

2.     Is this a highly conserved protein or DNA sequence?

3.     Are the sequences  used to make the alignment of roughly equal length? Explain.

4.     Do you see any significant shared insertions or deletions among phylogenetically related sequences

5.     For the actin-nuc data set only, does the nucleotide alignment correspond to the aa alignment?


 

3 & 4) Phylogenetic analysis.  We will explore the use of Phylip  and two different phylogenetic reconstruction methods (Distance and Parsimony). Phylip is free and available for many platforms, but it is a bit tedious to use.

 

Documentation for each of the phylip programs can be found in the documentation folder located with the Phylip programs.  Ask the course assistants to find out where these are located.  The documentation file names for the phylip programs are "seqboot.doc" or "neighbor.doc" etc...

 

Phylogenetic analysis with Phylip (Phylip is a collection of approx 30 programs).  We will only use 7 of them for this exercise.

 

Protdist - for generating a distance matrix with protein sequence

Protpars -  for generating parsimony trees with protein sequence

Dnadist -  for generating a distance matrix with DNA sequences

Dnapars -  for generating parsimony trees with DNA sequences

Neighbor - for generating either a neighbor-joining or UPGMA tree from a specified distance matrix (DNA or protein)

Seqboot  - for generating bootstrapped data sets (any type of sequence)

Consense  - for generating consensus trees

LOOK AT THE  DESCRIPTIONS OF EACH PROGRAM LOCATED BELOW FOR  MORE INFORMATION

 

As demonstrated in the workshop, you will need to pass your multiple sequence alignment through a series of programs. Each is step is quick. The programs are used in the following order:

Protdist  - neighbor      To construct protein distance tree

Dnadist  - neighbor       To construct DNA distance tree

Dnapars                       To construct DNA parsimony tree(s)

Protpars                       To construct Protein parsimony tree(s)

 

After completing the above and saving the data to files with new names, you can perform a bootstrap analysis of any of the above 4 scenarios, by first using seqboot (use 10 replicates for this exercise so that things will go quickly, but in real life choose at least 100 replicates), and in the end,  use consense on your treefile. e.g. seqboot -  protdist -  neighbor - consense.
Don''t forget to select the multiple option on each program in a bootstrap analysis, it must know how many data sets to expect.

 

The input for the first (or in the case of parsimony the only) program in each series is always your alignment in phylip format (file.phy renamed as infile).  ALL programs read "infile" whatever that may be (either your alignment or the output from an earlier step) except for consense, it reads treefile.


FIgure of paths 


All of your trees (treefile) can be viewed and edited in "treetool" and printed from there  ( NOTE: Treetool may not be installed, you may need to get and install the program.  How would you do that?.

 

Remember! Each phylip program only accepts infile and produces only outfile and/or treefile. Keep copies of each of these files with different names before moving on to the next step or you will have to begin again!. For example. My multiple sequence alignment was rpl2.phy


As an example I did the following:

 

cp cs.phy infile            (makes a copy of your alignment to use for phylip)

protdist                     (Makes protein distance calculations from infile)

cp outfile  protdist.out     (save a copy of output)

cp outfile infile            (copy output into infile for next program)

neighbor                     (makes tree from distances)

cp outfile neighbor.out      (save a copy of output from neighbor)  neighbor will also create a treefile

 

In the end, all you will need to keep is the file cs.phy (the alignment file) protdist.out (the distance matrix) neighbor.out (your branch lengths and groups) and treefile (your tree).

 

Look at the outfile generated at each of these steps. It will soon become very clear to you what is happening at each step. Also, play with the parameters offered in each program it only takes two minutes to see the results. What happens if you use a  matrix other than PAM for protdist?

 

 

You can download the programs or source code or documentation from http://evolution.genetics.washington.edu/phylip.html if you are interested.

 


Successful completion of this exercise requires that you generate and think about  the following;

 

        Three multiple sequence alignments: one for CS and two for Actin (one DNA and one Protein)

1.     The answers to the questions about the alignments listed above (a-d or a-e)


For the cs-prot data set, perform the following analyses:

1.     Distance (neighbor-joining) analysis with Phylip

2.     Parsimony analysis with Phylip

   
For the actin-prot data set, perform the following analyses:

1.     Distance (neighbor-joining) analysis with Phylip

2.     Bootstrap of the above analysis with Phylip


For the actin-nuc data set, perform the following analyses:

1.     Distance (neighbor-joining) analysis with Phylip

2.     Parsimony analysis with Phylip

 

For EACH tree, note the method(s) and relevant distance matrices or TS/TV ratios used to generate them (perhaps in a README file saved in the same directory as the trees)

 

      Think about the following:

1.     Did the different phylogenetic methods that you tried (distance and parsimony) yield consistent results when used on the same MSA?  Did you get exactly the same tree each time?

2.     If you look at the actin protein trees, what do they tell you about the evolution of actin?

3.     If you compare your DNA and Protein trees for actin, do they give the same trees?  If not why you think the results may have differed.

4.     Could you tell, just by looking at your multiple sequence alignments roughly how the trees would turn out?  What did you see that gave you your hunch or idea about the relationships?  Was your hunch correct?

5.     Turning from molecular systematics to molecular evolution, what interesting things happened in the evolution of the circumsporozoite protein?


EXTRA INFORMATION


PHYLIP is a collection of programs, each of which performs a special task.  You must call a new program for each task you want to perform. The programs relevant to this exercise are: seqboot, dnadist, protdist, neighbor, dnapars, protpars, consense. Each of these programs only accepts a file called infile by default (you can override) and each produces a file called outfile and/or treefile.  WARNING: Each of these programs will overwrite any existing outfiles or treefiles present in the same directory!

Seqboot makes bootstrapped datasets (i.e. alignments) that can be used in ANY analysis.  You must specify the number of datasets to make.  You only need this program if you are performing a bootstrap analysis.  It will create a single outfile.

Dnadist and Protdist are used to make distance matrices from DNA and protein data respectively. You must choose the weighting scheme and whether or not the data are interleaved.  If your infile happens to be the output of seqboot, you need to select the option indicating that multiple datasets are in use.

Neighbor is one of many programs that will make a tree from the data matrices created by dnadist and protdist. This program will create both an outfile and a treefile.  If you are running neighbor on a distance matrix file that originated from a seqboot set, i.e. a bootstrap, you need to select the option indicating that multiple datasets are being used.

Consense is only used on the treefile generated at the end of a bootstrap analysis.  It will compute a consensus tree from your bootstrap trees.  The input for consense is the treefile output generated by either dnadist, protdist, dnapars, or protpars.

Dnapars and Protpars perform parsimony analyses on dna and protein data respectively. These programs use the .phy multiple sequence alignment file as their infile..  If you need a parsimony bootstrap you must use seqboot first.  Dnapars and protpars will generate an outfile and a treefile.  If there are multiple trees or if you are doing a bootstrap analysis, you will need to use consense to get a consensus tree.

Once you have a treefile with a single tree in it, you can view that tree and edit it in treetool.  You can launch the program by typing “treetool”.  You can print you trees from treetool in many formats. You can also change the outgroup at this stage if you didn’t specify it correctly earlier and you can re-label branches with taxonomic names.

Examples: 
If you want to perform a distance analysis on a DNA data set you would do the following: dnadist then neighbor then treetool

If you want to perform a bootstrapped distance analysis on a protein data set you would do the following: seqboot then protdist then neighbor then consense then treetool.

A bootstrapped parsimony analysis on a protein data set  would look like: seqboot then protpars then consense then treetool.