I Bootstrap

Aim: To evaluate the level of support for internal branches in a phylogenetic
tree.  We shall demonstrate the differences we might see in "good" and "poor"
data sets.


We use a resampling procedure such as bootstrapping in an effort to evaluate
the relative levels of support and conflict for various sets of relationships. 
The original data is sampled with replacement and pseudo-replicates of the
original matrix are used to construct trees.  This resampling strategy is
repeated a large number of times and at the end the data is summarised using a
majority-rule consensus procedure.

1. Read the HIV.nex dataset into paup's memory.  Take a look at the bootstrap
commands options.  The first option is BSeed.  This is a seed number that
computers sometimes need in order to generate a random number.  PAUP will
usually use the clock time as a random number.

2. Use the 'bootstrap' command to perform a bootstrap analysis using the default
parameters.  Record the random 'starting seed' number.  Draw the resulting tree
into your lab book with the support values associated with the internal
branches.

3. You will also see a 'partition table' that details the number of times a
particular internal branch was seen during the resampling procedure.  you must
remember that during bootstrap resampling, it is likely that trees will be
produced that do not look like the final consensus tree.

This is an example of a partition table:

      12345678      Freq
      ------------------
      .***....     79.18
      ..**....     76.76
      .******.     72.31
      .....**.     52.28
      ....***.     50.51
      ....****     23.69
      .**.....     16.47
      .***.**.     15.65
      .*..***.      6.43


This indicates that during the analysis a branch that separated sequences
1,5,6,7 and 8 from the rest was seen 79.18% of the time.  Can you work out how
many times a branch was seen that separated sequences 2 and 3 from the rest? 
Would this branch be seen in the majority-rule consensus tree?

*NOTE:

The following notations can be used to describe the same tree:

Tree diagram -
                  ,>>>>>>>>>>>1:Cow
                  !  
                --6        ,>>2:Mouse
                  !  ,>>>>>8  
                  !  !     `>>3:Rat
                  `>>7  
                     !     ,>>4:Chimp
                     `>>>>>9  
                           `>>5:Human

Nested parentheses -
                ((Human, Chimp),(Mouse, Rat), Cow);


Partition table -
                   12345
                   ...**
                   .**..
                   .****


4. Read the Random.nex file into paup's memory.  Perform the same bootstrap
analysis on this data set.  Record the results and indicate how the trees differ
in terms of the degree of support seen throughout the tree.  Bear in mind that
the Random.nex file contains sequences that we have randomised - there is no
real set of phylogenetic relationships.

5. Perform the bootstrap analyses again on both datasets, this time alter the
number of replicates so that you perform 1,000 resampling iterations.

6. Record the results and quit the program.






