Molecular phylogenetic tree.

In this section you will plot a dendrogram displaying the measured similarities between the seven proteins which you pairwise aligned in Section 5. You will compare there relative position in the dendrogram to their relative position in the phylogenetic tree of life.

Below is the pairwise alignment scores from the 21 pairs aligned in section 5. The information in ~/tbss.work/Bioinformatics/multipleData/stats is assembled into a symmetric matrix:

	1EQR	1ATI	1ADJ	1EFW	1ASZ	1B8A	1BBW
1EQR	0.0	0.21	0.23	0.55	0.28	0.31	0.31
1ATI	0.21	0.0	0.25	0.24	0.24	0.18	0.21
1ADJ	0.23	0.25	0.0	0.24	0.24	0.21	0.23
1EFW	0.55	0.24	0.24	0.0	0.34	0.36	0.30
1ASZ	0.28	0.24	0.24	0.34	0.0	0.41	0.27
1B8A	0.31	0.18	0.21	0.36	0.41	0.0	0.28
1BBW	0.31	0.21	0.23	0.30	0.27	0.28	0.0

The commands in the following Matlab session are all in the Matlab script Dendro.m. The commands can be run all at once simply by typing Dendro at the Matlab command line when Matlab's current directory contains Dendro.m.

Now we will use the clustering algorithms in the Statistics toolbox of Matlab to draw a dendrogram of the relatedness of the domains. Here we use the above scores derived from sequence alignment, but structure alignment scores could be used as well[4].

• Move to the directory for this exercise with cd ~/tbss.work/Bioinformatics/matlabData

• Start Matlab by typing at the UNIX console: matlab.

• The commands in the following Matlab session are all in the Matlab script Dendro.m. The commands can be run all at once simply by typing Dendro at the Matlab command line, as long as Matlab's current directory contains Dendro.m and distM.dat. If you like, type in the below, or paste lines into the Matlab command line from Dendro.m or the web-based version of this tutorial. (To see the numerical result of a calculation, leave of the semicolon from the end of the line. To see the value of a variable, enter its name alone on the Matlab command line.)

• First, we read in the above distance matrix of sequence similarity for 7 proteins.

  load distM.dat;
  

• We make a new matrix by subtracting the sequence similarity values from 1, so that longer distances in our dendrogram will correspond to greater evolutionary distance.

  dM=1-distM;
  

• Its important to keep track of names of the proteins...

   
  l={'1eqr','1ati','1adj','1efw','1asz','1b8a','1bbw'};
  

• To use the 'linkage' command of Matlab, one needs to form a column vector of the $( (n) (n-1) / 2 )$ non-redundant elements above the main diagonal of the $n \times n$ distance matrix; our $7 \times 7$ matrix produces a 21-element vector:

  d=[dM(2:7,1);dM(3:7,2);dM(4:7,3);dM(5:7,4);dM(6:7,5);dM(7:7,6)];
  

• Use the linkage command to make a hierarchical cluster tree using average distance between cluster elements:

  z1=linkage(d','average');
  

• For more options in constructing the cluster tree, type help linkage at the Matlab command line, also see a modeling text such as Leach [10].

• We display the dendrogram of the clusters in z1:

  h101=figure(101);
  dendrogram(z1);
  

• And, finally, paste in some magic to place the labels correctly:

   
  hx=get(get(h101,'CurrentAxes'),'XTickLabel');
  for i=1:size(hx,1)
     hx(i)=str2double(hx(i));
  end
  set(get(h101,'CurrentAxes'),'XTickLabel',[l(hx(1)), ...
           l(hx(2)),l(hx(3)),l(hx(4)),l(hx(5)),l(hx(6)),l(hx(7))])
  figure(h101);
  title('Molecular Phylogenetic Tree');
  xlabel('Protein (pdb code)')
  ylabel('1-Similarity (%)')
  

• Print out the dendrogram, or copy it down on paper, along with the names of the proteins. Refer to Table 6 to write, under each name, the domain of life each protein originates from.