Ancestral reconstruction of protein sequence

[zaryanichka]

I received very strange result of ancestral reconstruction.

1. In the situation where all of the sequences in file given have the same amino acid in one position in the ancestral state in this position another amino acid has the greatest probability.
2. Considered some range of probabilities. First letter is M everywhere, but with a very close probability it is G.

I've started calculation strictly according to the instruction. This is also not consistent with the replacement matrix used(blosum).

[nylander]

We would be helped with some example data set, and commands issued, in order to reproduce this behavior.