Skip to main content
. 2018 Aug 23;7:e36317. doi: 10.7554/eLife.36317

Figure 2. BGS and gBGC both have an impact on the SFS.

Each panel corresponds to the normalized unfolded SFS of Yoruba (top, YRI) and Japanese (bottom, JPT) populations. (A) SFS computed for all SNPs in the 2nd, 10th and 20th recombination classes (as defined in Figure 1). For each panel, pairwise comparisons of the SFS are significant with p-values<10–3 (see Materials and methods). The SFS for all ten 1000G populations are shown in Figure 2—figure supplement 1. (B) SFS for three gBGC mutation categories computed for three recombination classes. Note that WW and SS sites (in red) are unaffected by gBGC. All SFS are different from each other (site permutation test, p-values<10–3) except for the Yoruba recombination class two between WS and WWSS where p=0.0135. (C) SFS for sites unaffected by BGS and gBGC (WW + SS sites with RR ≥1.5 cM/Mb). The four SFSs are not significantly different from each other at the 1% significance level, as revealed by a permutation approach (see Materials and methods). Shaded areas delimit 95% confidence intervals using a block-bootstrap strategy (see Materials and methods).

Figure 2.

Figure 2—figure supplement 1. SFS of ten 1000G populations for sites belonging to three recombination classes.

Figure 2—figure supplement 1.

Recombination classes 2 (low), 10 (medium), and 20 (high) correspond to the respective recombination bins as defined in the main text. (A) Normalized unfolded SFS, (B) unfolded SFS. Low-recombination classes show increased singleton frequencies and a lower proportion of intermediate- and high-frequency variants. For each panel, pairwise comparisons of the SFS are significant with p-values<10–3 (see Materials and methods). Shaded areas correspond to 95% CI obtained by a block-bootstrap approach (see Materials and methods).
Figure 2—figure supplement 2. (A) Normalized unfolded SFS of ten 1000G populations for sites belonging to three recombination classes in NTR regions more than 50 kb away from transcribed regions (NTR-50kb sites).

Figure 2—figure supplement 2.

(B) Zoom on singletons and doubletons categories. The singletons show a significantly increased frequency in the low-recombination class, in keeping with the action of BGS in non-coding regions. For each panel, pairwise comparisons of the three SFS are significant with p-values<10–3 (see Materials and methods). Shaded areas (panel A) and segments (panel B) correspond to 95% CI obtained by a block-bootstrap approach (see Materials and methods).
Figure 2—figure supplement 3. Impact of covariates on the normalized unfolded SFS for WW + SS sites with RR ≥ 1.5 cM/Mb (in red) on Yoruba (top) and Japan (bottom) population.

Figure 2—figure supplement 3.

(A) Recombination distance to hotspots (i.e. RR > 10 cM/Mb). SFS of sites belonging to the first (d ≤ 0.01647 cM, in green) and fourth quartile (d ≥ 0.08728 cM, in blue) are significantly different from the SFS computed on all SNPs (p=0.017 and p=0.005 respectively for Yoruba, p<10−3 for Japan, see Materials and methods). (B) Recombination distance to PhastCons conserved elements (see Materials and methods for definition). SFS of sites belonging to the first (d ≤ 0.0000543 cM, in green) and fourth quartile (d ≥ 0.0024580 cM, in blue) are significantly different from the SFS computed all SNPs (p<10−3 for all but the 4th quartile of Yoruba population for which p=0.003). (C) The SFS of exonic sites (green) are significantly different from that computed on all SNPs (p=0.004 and p<10−3 for Yoruba and Japan resp.) while the SFS of non-exonic sites (blue) are not different from the SFS of all SNPs (p=1). The SNPs furthest to hotspots have both an enrichment for singletons and high-frequent variants, whereas the SNPs closest to conserved elements and exonic sites have only an enrichment for singletons. Shaded areas correspond to 95% CI obtained by a block-bootstrap approach (see Materials and methods).
Figure 2—figure supplement 4. Impact of physical distance to phastCons conserved elements on the normalized SFS for WW + SS sites with RR ≥ 1.5 cM/Mb (in red) on Yoruba (left) and Japan (right) populations.

Figure 2—figure supplement 4.

Sites that are less (resp. more) than 0.0003 cM from conserved elements are in green (resp. in blue). phastCons elements are defined in the Materials and methods section. The SFS of sites with d > 0.0003 cM is not different from all sites for both populations while the SFS of sites with d ≤ 0.0003 cM is (p<10−3). Shaded areas correspond to 95% CI obtained by a block-bootstrap approach (see Materials and methods).