Figure 1.
Relationship between substitution rate and heterozygosity among 40 humans from 10 populations.
Populations are: LWK (red), MKK (black), YRI (white), TSI (green), CHB (grey), JPT (dark blue), UTA (yellow), ASW (light blue), GUJ (orange), MXL (purple). Abbreviations are given in methods. Mean relative substitution rate is calculated for each individual relative to all others, negative scores indicating lower than average and positive indicating above average, and expressed as substitutions per qualifying base. Heterozygosity is calculated as the proportion of heterozygous sites among all qualifying bases. Both axes are scaled for clarity. Standard errors for each data point are nominally of the order of 0.035, making them too small to show clearly as error bars.
Figure 2.
Genetic divergence of 40 humans from the chimpanzee reference sequence according to genomic region.
Divergence rates are quantified as the proportion of qualifying bases that differ with no adjustment for differences in rate of transition and transversions. Populations are: Africans sampled in Africa (black, ordered from left LWK, MKK, YRI), Africans sampled in America (ASW, grey) and non-Africans (open circles, four of each ordered from left TOS, CHB, JPT, UTA, GUJ, MXL). Panels are: all autosomes (Au); a randomly-selected medium sized autosome, chromosome 9 to compare with the X (C9); the X chromosome (X); the Y chromosome (Y, only half the samples are males). All autosomes yield very similar patterns with all Africans showing higher divergence than all non-Africans.
Figure 3.
Degree of ‘out of Africa’ loss of heterozygosity predicts African – non-African substitution rate difference across 1272 genomic regions.
The genome was divided into ∼2 Mb non-overlapping windows and, within each, values derived for the average heterozygosity and average substitution rate in 12 non-admixed Africans (LWK, MKK, YRI) and 12 non-admixed non-Africans (TSI, CHB, JPT) yielding paired difference values. The raw data exhibit considerable scatter. For clarity, data were binned by size of heterozygosity difference and both axes were rescaled for clarity. Three data points are omitted where partial windows at the end of a chromosome yielded appreciably fewer qualifying bases. The third order polynomial fitted to the means is highly significant (r2 = 0.993, t = 39.54, 11 d.f., P = 3.3×10−13) as are linear and polynomial regressions fitted to the raw data (see text). Changes in heterozygosity are dominated by the ‘out of Africa’ bottleneck, modulated by natural selection, hence are almost invariable positive (heterozygosity greater in Africa).
Figure 4.
Average difference in heterozygosity between closely related populations in regions surrounding putative new mutations (PNMs).
PNMs are defined as derived variants that occur only once among 80 human chomosomes studied. The x-axis is heterozygosity in a less related outgroup population, quantified as total number of heterozyous sites in the 2 Kb window summed over all four individuals, while the y-axis is the difference in heterozygosity between the population in which the PNM was inferred and a closely related sister population, measured as average excess sites per individual. Reciprocal comparisons were conducted between two pairs of sister populations: Uta (open squares) and Tos (black squares) in Europe and Chb (grey crosses) and Jpt (grey circles) in east Asia. Outgroups were Jpt and Tos four Europe and East Asia respectively. In all cases, heterozygosity in the population in which the PNM was inferred is on average greater than in the control populaiton, this difference rising as heterozygosity increases in the outgroup. Results presented are weighted linearly by proximity to the putative new mutation, but an unweighted analysis yields essentially identical results.