Figure 4.

Genealogy for five sampled sequences illustrating how closed intervals, |s_i|, and Σ_i|s_i| are calculated by the KM (a) and KM' (b) models. The closed and open segments of each bar denote the available versus missing or ignored regions of its sequence as scored over the range of (0:1). |s_i| are calculated from these scores (numbers next to branches) and Σ_i |s_i| is then determined as their sum for each T_j (values on the far right). In contrast, the KM' model (b) rescores the sampled sequences and the (n - 1) basal ancestral sequence of this asymmetrical genealogy (the diamond) as ∅. Thus, these sequences make no contribution to Σ_i |s_i| as their |s_i| = 0.0. The KM' model also ignores the first 20% of the sequence for the common ancestor of the two leftmost sampled sequences. This region is missing from the leftmost sampled allele and thereby fails the "below" half of the "above and below" test (see text).