Biology Reference
In-Depth Information
Exercise 10.18.
1. Apply the NJ Algorithm to the dissimilarity map D specified by the information
given below. (Recall that any element marked
can be inferred from the data
given.) Take note of the order in which cherries are picked, and when more than
one choice of cherry could be made. 3
54768
∗∗
7 09 1
∗∗∗
768
∗∗∗ ∗
D
=
59
∗∗∗∗∗
8
∗∗∗∗∗∗
2. The software package BIONJ is an improved version of the NJ Algorithm that
generally produces the same trees when the number of leaves is small. Access
BIONJ at http://www.phylogeny.fr , e.g., from the homepage under
“Phylogeny” pop-up, located under the heading “Online Programs” in the bar
near the top of the page.
a. First, click on the option to “load an example of a distance matrix.” Scroll
down and click the “Submit” button to see the output. Click on the “Data and
Settings” tab above the text box to return to the input dissimilarity map.
b. Based on that experience, now replace the example data with the correspond-
ing distancematrix data for D in part (1) of this problem, adjusted accordingly,
and apply BIONJ. Use the leaf names “A, B, C, D, E, F,” in that order, with
A corresponding to the first row/column of the dissimilarity map D , and so
on. If Java is supported, the button “Visualize your tree with ATV” can also
be used for another view of the tree.
c. For more practice with real species:
i. Go to http://www.atgc-montpellier.fr/fastme/
fastme_ex.txt Three dissimilarity maps appear there. Try loading
the first one into BIONJ. (Note: For implementation at http://www.
phylogeny.fr you will need to put the species name and the corre-
sponding numerical data for its row together on the same line.)
ii. Predict, from the output, the first cherry that would be picked by the NJ
Algorithm, and then verify your guess by computing the Q -criterion for
the dissimilarity map (i.e., carry out the first step in the NJ Algorithm).
iii. What species are being related, here? You may wish to make use of the
search feature at http://tolweb.org/tree/ .
3 Though tedious, carrying out these computations will provide the experience necessary to make sense
of the algorithm. A tutorial for the first step, using an equivalent arithmetic formulation of the Q -criterion
presented here, is given at http://www.icp.ucl.ac.be/˜opperd/private/neighbor
html .
 
Search WWH ::




Custom Search