Table 3.
(∆L A, ∆L D) | Ratio (= ref/KS) a | Overlapping b |
---|---|---|
(1, 1) | 1.667 | 0.167 |
(3, 1) | 1.883 | 0.383 |
(3, 3) | 2.449 | 0.949 |
(5, 5) | 3.325 | 1.825 |
(10, 1) | 2.165 | 0.665 |
(10, 10) | 5.572 | 4.072 |
(25, 1) | 2.355 | 0.855 |
(25, 4) | 4.714 | 3.214 |
(30, 10) | 8.300 | 6.800 |
(100, 1) | 2.561 | 1.061 |
(100, 3) | 4.896 | 3.396 |
(300, 1) | 2.659 | 1.159 |
Each row gives values for a local PWA with ∆L A ancestral sites and ∆L D descendant sites in between a pair of PASs. See section M1 of Methods for the parameter setting. When λ I = λ D, the ratio for (ΔL A, ΔL D) = (L 1, L 2) is identical to that for (ΔL A, ΔL D) = (L 2, L 1). Thus, we only showed the results for ΔL A ≥ ΔL D. The ratios less than 2 are in boldface. This table is a modified version of Table 4 of [43]
a The ratio of the ab initio probability to the corresponding probability given by the HMM of Kim and Sinha [36], in the limit where the time interval (or branch length) approaches zero
b The portion of the ratio contributed by overlapping indels (such as those in panels d and e of Additional file 1: Figure S3)