According @glipsnort’s article at 32, the sum of G-C, A-T, A-G/G-T mutations should be roughly equal to the number of transitions mutations (C-T or A-G). So, in your example, on the 22 cases of homoplasy, we would have expected to see about 50% of transitions, not 100% as it is the case. What is the explanation for this strange observation?