The DAG allows one to clearly read the dependencies from the model, and makes it easier to write down the likelihood. The samples from the posterior distribution are drawn using the numerical techniques already discussed.
