Contrasting signals of positive selection in genes involved in human skin-color variation from tests based on SNP scans and resequencing

Table 4 Median and 95% Credible Interval (CI) of the estimates of the time when selection started

Gene	Population under positive selection^a	Time of selection (95% CI)	σ (95% CI)	h(95% CI)
TYRP1	Europe^b	50,250 (13,500 to 97,250)	361.94 (55.6 to 984.79)	1.70 (0.71 to 4.28)
OCA2	Europe,^b East Asia	45,875 (12,750 to 95,006)	383.42 (36.07 to 992.04)	2.04 (0.78 to 4.56)
KITLG	Europe,^b East Asia^b	46,000 (11,750 to 93,269)	385.49 (37.68 to 1036.32)	2.01 (0.74 to 4.41)
DCT	East Asia^b	48,500 (11,250 to 97,006)	789.24 (55.21 to 1584.98)	1.90 (0.67 to 4.11)
SLC45A2	Europe^b	27,500 (10,500 to 73,294)	676.53 (127.73 to 1026.06)	1.10 (0.53 to 2.60)
Neutral simulation	Europe^b	44,875 (12500 to 95,756.25)	593.18 (47.99 to 1086.22)	1.69 (0.56 to 3.97)

^aThe table shows the selection parameter σ (defined as 4 × N_e × s, where N_e is the harmonic mean of the effective population size through time) and overdominance parameter h computed for each population that has been suggested as being under selective pressures in the genes that have been sequenced, data from Soejima et al. [30] for SLC45A2, known to be under selective pressure in Europeans (see Methods), and 5 kb produced under a neutral out-of-Africa model (s = 0, h = 0).
^aFrom Lao et al. 2007. Prior distributions (without scaling) were truncated normal (mean ± sd 1810 ± 800 generations, range 400 to 4000) for the time of selection, truncated normal (mean ± sd -4 ± 3, range -4 to 2) for s in logit form, and truncated normal (mean ± sd 1 ± 1, range 0.5 to 6) for h. In total, 2,000 simulations were performed for each gene using as a prior distribution of the demographic parameters the median value of the posterior distributions obtained from the 50 loci (see Methods). A burn-in of 1,000 simulations was used.
^bPopulation where the selective pressure was simulated.

ISSN: 2041-2223