Abstract: For a set of sequences, related by a binary tree, that have evolved according to the Thorne-Kishino-Felsenstein model, a Gibbs sampler is presented for simulating the ancestral sequences and their alignments. The updating step consists in updating the ancestral sequence and its three alignments within a 3-star tree. We compare the Gibbs sampler with the algorithm suggested recently by Holmes and Bruno.
Key words and phrases: 3-star tree, EM-algorithm, evolutionary model, hidden Markov model, Markov chain Monte Carlo.