This is somewhat a cheat - I'm not actually writing the tutorial, but dropping a link to an already excellent tutorial written by Trevor Bedford. That tutorial gives a demo of a whole analysis pipeline from
BEAUTI through to
TreeAnnotator. Ultimately, though, the more difficult question is about the most appropriate choice of parameters to select for the given study.
This set of Taming-the-Beast(2) tutorials are also useful, though applicable only to BEAST 2.
If I read the documentation correctly, BEAST provides some statistical support for the quality of the tree by recomputing the tree over and over again from the same input - they tend to call this "bootstrap support". This is not called bootstrap, it's called Monte Carlo. Bootstrap in statistics strictly refers to sampling with replacement. A Monte Carlo simulation is a procedure where, say, you wish to find out the probability of heads on a weighted coin (so not necessarily 0.5). Flip the coin 1000 times; the proportion of the number of times that the coin comes up heads will be the empirical probability of heads. That's a Monte Carlo simulation.