Bootstrapping involves taking many randomized subsamples of your data, performing the same statistic or function on each subsample, and then analyzing the distribution of results. For example, if you want confidence intervals on the median value of some distribution, you should take randomized subsamples of the population, compute and store the median of the subpopulation, and repeat the process many times until you have a distribution of median values. In that example, @fun would be @median.
Let's say your distance matrix is computed by your homemade function,
output = computePhyloDistance(input)
then your @fun would be @computePhyloDistance.
Step through some of the examples provided by Matlab and the process will become clear.
