Human and non-human primate genomes share hotspots of positive selection

Abstract

Among primates, genome-wide analysis of recent positive selection is currently limited to the human species because it requires extensive sampling of genotypic data from many individuals. The extent to which genes positively selected in human also present adaptive changes in other primates therefore remains unknown. This question is important because a gene that has been positively selected independently in the human and in other primate lineages may be less likely to be involved in human specific phenotypic changes such as dietary habits or cognitive abilities. To answer this question, we analysed heterozygous Single Nucleotide Polymorphisms (SNPs) in the genomes of single human, chimpanzee, orangutan, and macaque individuals using a new method aiming to identify selective sweeps genome-wide. We found an unexpectedly high number of orthologous genes exhibiting signatures of a selective sweep simultaneously in several primate species, suggesting the presence of hotspots of positive selection. A similar significant excess is evident when comparing genes positively selected during recent human evolution with genes subjected to positive selection in their coding sequence in other primate lineages and identified using a different test. These findings are further supported by comparing several published human genome scans for positive selection with our findings in non-human primate genomes. We thus provide extensive evidence that the co-occurrence of positive selection in humans and in other primates at the same genetic loci can be measured with only four species, an indication that it may be a widespread phenomenon. The identification of positive selection in humans alongside other primates is a powerful tool to outline those genes that were selected uniquely during recent human evolution.

Figure 1. Known selective sweeps identified in individual human genomes.

From left to right, each graph presents the variation of K at four known examples of selective sweeps in the European human population, namely the lactase gene LCT, FOXP2, OCA2-HERC2 and SLC24A5. The upper, middle and lower panels show genome scans (200 kb windows in steps of 10 kb) for CV, JW and the average between both individuals, respectively. The X axis indicates chromosome coordinates in megabases, the Y axis indicates the log₂ of K (+0.001 to avoid null values). Dotted horizontal lines delineate the K = 0.05 threshold. Genes are highlighted in grey above the X axis, with the gene known to be affected by positive selection highlighted in black.

Figure 2. Human and other primate orthologous genes co-occur in selective sweeps more often than expected by chance.

The relative co-occurrence excess (observed score/expected score –1) between human and other primate candidates was measured for several upper thresholds of K ranging from 0 to 0.1. *: P≤5.10⁻². **: P≤10⁻². ***: P≤10⁻³.

Figure 3. Orthologous genes co-occur in selective sweeps more often than expected by chance.

There are 61 triplets and 11 quartets of orthologous genes that occur in putative selective sweeps simultaneously yet independently in human and two or three other primate genomes respectively (C₃+C₄ score = 227, black arrow). This is significantly more than expected by chance as shown by a co-occurrence test (black distribution) with 100,000 iterations, even after controlling for gene density alone (blue distribution), recombination alone (red distribution), or gene density and recombination simultaneously (green distribution) (see Methods). The dotted black distribution represents the hypothetical distribution where the observed score would still be significant at the 5% threshold.

Figure 4. Recent human versus coding sequence positive selection.

The co-occurrence test was used between coding sequence positive selection detected with the test 2 of PAML separately in five primate phylogenetic branches one the one hand, and recent positive selection in human and at least one additional primate one the other hand (see Methods; Figure S5). Genes were considered as positively selected in specific branches if twice the log-likelihood ratio (2ΔL) obtained with test 2 was greater than an arbitrary threshold comprised between 20 and 100. *: P≤5.10⁻². **: P≤10⁻². ***: P≤10⁻³.

Figure 5. Co-occurrence between non-human primates selective sweep candidates and XP–EHH worldwide human populations selective sweep candidates.

The co-occurrence test was applied using K in chimpanzee, orangutan and macaque to assign candidate genes on the one hand (Figure S5, group 1) and increasing XP-EHH values at genomic centers of genes tested in seven human populations on the other hand (Figure S5, group 2). The histogram represents the relative co-occurrence excess obtained using 9,873 orthologous genes (observed score/expected score – 1) for XP-EHH values increasing from 2 to 3.5 with recombination and gene density being accounted for (see Methods). The excess of co-occurrence observed is lower for K≤0.01 than for K≤0.05, likely reflecting a loss of power to detect hotspots at the most stringent thresholds. *: co-occurrence test P≤0.05. **: P≤0.01. ***: P≤0.001.

Figure 6. Co-occurrence between PAML branch-site test 2 positive selection candidates in non-human branches and XP-EHH worldwide human populations selective sweep candidates.

The co-occurrence test was applied using increasing 2ΔL values (twice the log-likelihood ratio) to assign positive selection candidate genes in chimpanzee, orangutan, macaque and human-chimpanzee branches of a primate phylogenetic tree on the one hand (see Methods; Figure S5, group 1), and increasing XP-EHH values at genomic centres of genes tested in seven human populations (Figure S5, group 2) on the other hand. The histogram represents the relative co-occurrence excess (observed/expected – 1) for combinations of XP-EHH threshold values increasing from 2 to 3.5 and branch-site test 2 2ΔL thresholds from 20 to 100 (Methods). *: co-occurrence test P≤0.05. **: P≤0.01. ***: P≤0.001.

Figure 7. Candidate hotspots of recent positive selection at the Toll-like receptors 1, 6, and 10 cluster and the FOXP2 locus.

Each graph shows the variation of the log₂ of K (+0.001 to avoid null values) at two candidate hotspots of recent adaptive evolution in human (average of the two individuals, red), chimpanzee (blue) and orangutan (green). To facilitate comparisons between genomes, values of K for chimpanzee and orangutan were projected on their human orthologous coordinates and gene symbols are those for human in all three species. Other legends are identical to Figure 1.