Human chromosomes are made up of DNA, which contains about 3 billion ‘letters’ that carry the instructions needed to build and maintain an individual. However, only about 10 percent of the human genome is made up of genes that code for proteins, or have a defined role in the body. The DNA sequence is largely the same in all people, but some modifications – or variants – occur about every hundred letters. These produce different versions of the same gene, which give us our unique features, such as the color of our hair or eyes.

The frequencies of some genetic variants can change over time, which makes human populations diverge genetically and physically. This can happen through different mechanisms. Positive selection keeps variants that are beneficial in specific environments, while negative selection removes genetic changes that are detrimental, for example because they cause disease. Transmission bias favors one of the two variants from our two parents. Chance alters the frequencies of neutral variants, which are neither good nor bad for the individual.

It is important to distinguish between these different scenarios, as they inform us about the forces that act on human evolution. For example, neutral variants tell us about the demography and migration patterns between populations. Variants under negative selection reveal which genetic areas are under pressure to stay the same because they are important for the organism to function correctly. Until now, it was unclear how we could best identify the variants affected by different evolutionary pressures, and how much of the genome was under negative selection.

Pouyet, Aeschbacher et al. created a measure of genetic diversity that is only affected by selection or transmission bias. The results showed that negative selection influences as much as 85 percent of our genome, whereas transmission bias affects a majority of the rest of the genome. After removing these two biases, less than 5 percent of the human genome is found to evolve by chance. This suggests that while most of our genetic material is formed of non-functional sequences, the vast majority of it evolves indirectly under some type of selection.

These findings define which parts of our genome evolves neutrally and can therefore be used to correctly reconstruct the past demography and migration events of humans around the world. The next step could be to reassess the history of human populations that was drawn using genomic data.

Human genomic diversity has evolved under diverse and complex constraints (Auton et al., 2015), such as past demography, selection, mutations, or genomic rearrangements (Lohmueller et al., 2011; Schiffels and Durbin, 2014; Sudmant et al., 2015; Mallick et al., 2016). However, the influence of these evolutionary forces and their interactions remain to be fully understood. For instance, it is yet unclear which fraction of the genome evolves under positive or purifying selection (McVicker et al., 2009; Rands et al., 2014; Corbett-Detig et al., 2015). Such information is crucial to our understanding of what portion of the genome is evolving neutrally, and necessary to form a clear basis for demographic inference, the detection of selective events, or the inference of the distribution of fitness effects of new mutations.

Genome-wide variation in recombination may strongly affect neutral variants (Spencer et al., 2006; Corbett-Detig et al., 2015), as selection will have more impact on linked polymorphism in regions of low recombination (Charlesworth et al., 1995), whereas biased gene conversion, which can also mimic the effect of selection (Galtier and Duret, 2007; Ratnakumar et al., 2010), will occur mostly in regions of high recombination (Katzman et al., 2011). In humans, various measures of diversity are positively correlated with levels of recombination (Nachman, 2001; Spencer et al., 2006; Cai et al., 2009; Lohmueller et al., 2011). While a direct mutagenic effect of recombination seems unlikely (McVicker et al., 2009; Schaibley et al., 2013) except at CpG sites (Arbeithuber et al., 2015), there is still some debate about whether the correlation between diversity and recombination is driven by recurrent selective sweeps (hitchhiking of neutral and slightly deleterious mutations) or background selection (BGS; i.e. purifying selection against deleterious mutations at linked sites) (McVicker et al., 2009; Stephan, 2010; Hernandez et al., 2011; Lohmueller et al., 2011). The modeling of genomic diversity under selection in humans suggests that it can be explained entirely by BGS (Lohmueller et al., 2011), whereas a combination of both BGS and positive selection seems to best explain genomic diversity in Drosophila (Elyashiv et al., 2016). However, the correlation between diversity and recombination is generally relatively weak in humans for most tested statistics and seems restricted to genomic regions of relatively low-recombination rate (<1 cM/Mb, (Cai et al., 2009; Lohmueller et al., 2011)).

Given the positive relationship between recombination and genetic variability, it has been proposed that the genomic regions most suitable for demographic inferences should be far away from genes and have high-recombination rates (Lohmueller et al., 2011). However, regions of high recombination might be prone to GC-biased gene conversion (gBGC), a process by which GC alleles in recombination tracts are preferentially transmitted in GC/AT heterozygotes (Duret and Galtier, 2009). This process thus increases the frequency of G and C derived alleles (usually denoted as strong or S alleles, Lachance and Tishkoff, 2014) relative to A and T (denoted as weak or W alleles), especially in recombination hotspots (Spencer et al., 2006; Glémin et al., 2015). By modifying allele frequencies in high-recombination regions, gBGC affects the site frequency spectrum (SFS) (Lachance and Tishkoff, 2014; Glémin et al., 2015) such that it becomes right-shifted for W-to-S (WS) mutations and left-shifted for S-to-W (SW) mutations. In addition, gBGC affects various classical statistics used to detect selection, and WS SNPs show larger levels of population differentiation than other SNPs (Lachance and Tishkoff, 2014). Overall, gBGC is believed to directly affect only 1% to 2% of the human genome, near recombination hotspots (Glémin et al., 2015), but due to the transient nature of these hotspots, a larger fraction of the genome could have been affected in the long term.

Here, we use two whole-genome human datasets to determine how and to what extent recombination and selective forces affect genome-wide diversity in humans. We examine the relationship between recombination rate and the average derived allele frequency per individual, as well as the SFS. After determining the parts of the genome that are least affected by BGS and gBGC, we examine the impact of these two processes on the SFS, and how they affect demographic inference based on the SFS.

We first used a representative set of one hundred individuals from the 1000 Genomes (1000G) Project (Auton et al., 2015) from ten populations in five geographic regions to study the pattern of human genomic diversity. Since our analyses compared genomic diversity across individuals for sets of sites devoid of any missing data, we selected in each population those 1000G individuals with the highest coverage. As a measure of genomic diversity, we used the average derived allele frequency per individual (${\displaystyle \overline{DA{F}_i}}$). This statistic was computed over all sites that were found polymorphic across all populations (i.e. where derived alleles are neither fixed nor absent in all individuals). Assuming that there are S_Tot such sites, ${\displaystyle \overline{DA{F}_i}}$ is computed for each diploid individual as the total number of derived alleles observed at those sites (n_i) divided by 2 S_Tot. We show in the Materials and methods below how this statistic depends on the average time to the most recent common ancestor (t_MRCA) of the whole sample, and, if one assumes neutrality that this statistic should be the same on expectation for any individual in the sample across its whole genome, irrespective of the particular demography of its population (Figure 1—figure supplement 1). Differences in the number of derived alleles (n_i) among individuals and among genomic regions should therefore only reflect differences in selection, mutation rate and/or generation time (Figure 1—figure supplement 2). The number of derived alleles is indeed broadly comparable across individuals from different geographic regions, even though Southern and Eastern Asians (SAS and EAS) show a slight yet significant deficit in the number of derived alleles than the three other groups (~50,000 out of 17 million, Tukey test, p<0.01) (Figure 1—figure supplement 2), suggesting either a more efficient selection, a lower mutation rate or a longer generation time. This statistic thus appears ideally suited to assess the impact of selection at linked sites that could locally alter their t_MRCA.

Figure 1 with 10 supplements see all

Download asset Open asset

BGS and gBGC affect the whole SFS

Since ${\displaystyle \overline{DA{F}_i}}$ increases with recombination rate (Figure 1), BGS does not simply amount to lowering the effective population size (Charlesworth, 1994; Charlesworth et al., 1995; Hudson and Kaplan, 1995), as this simple rescaling would not modify allele frequencies. BGS thus affects the SFS (Zeng and Charlesworth, 2011) in complex ways (Nicolaisen and Desai, 2013), and the comparison of sites that are differentially exposed to BGS allows us to better examine this influence. The SFS computed in ten 1000G populations for different recombination classes (Figure 2A, Figure 2—figure supplement 1) shows distortions that are qualitatively similar in all populations, irrespective of differences in demographic history. As compared to the highest recombination class, the second-to-lowest recombination class (which is potentially the one most strongly affected by BGS) not only shows an excess of singletons, but also a deficit of intermediate and high frequency variants (Figure 2A). Similar distortions are also observed in non-transcribed regions, and even (but to a lower extent) in regions at least 50 kb away from transcribed regions (Figure 2—figure supplement 2), in line with our results for ${\displaystyle \overline{DA{F}_i}}$.

Figure 2 with 4 supplements see all

Download asset Open asset

To understand the respective effects of gBGC and BGS on the SFS, we computed the SFS for subsets of mutations differentially affected by gBGC in the Yoruba (YRI) and Japanese (JPT) 1000G populations (Figure 2B). In line with previous work (Lachance and Tishkoff, 2014), we find that the difference between the SFSs of unbiased mutations (WW + SS) and biased mutations (SW and WS) increases with recombination rate. In particular, WS mutations show a deficit of low-frequency variants and an excess of intermediate- and high-frequency variants in regions of high recombination (Figure 2B). As previously recognized (Katzman et al., 2011; Lachance and Tishkoff, 2014), the excess of high-frequency variants at WS sites is not compensated by a corresponding deficit of high-frequency variants at SW sites, implying that gBGC could contribute to the increase of nearly fixed derived alleles that has previously been attributed to mislabelled ancestral states or positive selection (Hernandez et al., 2007).

Impact on demographic inferences

To investigate the impact that the choice of SNPs may have on demographic inference, we estimated demographic parameters for the Yoruba and Japanese populations using three different SFSs (Figure 3A): the synonymous SFS commonly used in exome resequencing studies; the SFS inferred on non-transcribed regions at least 50 kb away from coding regions (NTR-50kb), and on our best-filtered dataset (WW + SS sites in ≥1.5 cM/Mb regions), hereafter called the ‘neutral’ dataset. Note that this neutral SFS was computed over both TR and NTR regions since they show the same SFS (Figure 2C). Interestingly, the SFS observed at synonymous sites differs markedly from that observed at neutral sites, as it comparatively shows a significant deficit of low-frequency variants and a large excess of high-frequency variants (Figure 3A, Figure 3—figure supplement 1A). It appears that this latter excess is due to gBGC, as it disappears when one computes the SFS on synonymous sites not affected by gBGC (Figure 3—figure supplement 1B).

Figure 3 with 3 supplements see all

Download asset Open asset

Using a simple demographic model of a focal population going through three successive bottlenecks and receiving some migrants from surrounding populations (modelled as a ghost population for simplicity) (Figure 3C), we can fit almost perfectly the three SFSs (Figure 3—figure supplement 2). Yet, the inferred parameters differ considerably (Supplementary file 3 - Table S3). For the Yoruba population, the differences in demography are especially important in the old periods (>100 ky, Figure 3B). With the neutral SFS, we nevertheless infer a more recent last bottleneck dated at the end of the Last Glacial Maximum (LGM), a more pronounced and more recent admixture event from surrounding populations. The ancient demography is markedly different with a significantly more ancient second bottleneck and a significantly lower ancient population size inferred from both synonymous and NTR-50 kb SFS. The Japanese demography inferred from the three data sets shows more similarity over the last 600 ky but the demography inferred from the neutral data set suggests a stronger recent bottleneck (pre LGM) and no population expansion as compared to what is inferred from the synonymous SFS neutral data set. Our results thus clearly show that very different demographies can be inferred from neutral and non-neutral SFSs. However, even though BGS and gBGC affect the SFS of populations with distinct histories in a qualitatively similar way, they have different consequences on their reconstructed demography. It thus appears difficult to predict how demographic parameters will be biased when using non-neutral SFS.

Simulations of BGS reproduce observed patterns

To confirm that our observed patterns were compatible with background selection, we ran individual-based forward simulations implementing BGS with SLiM v. 2.3 (Haller and Messer, 2017) in populations having the demography estimated from neutral sites in the Japanese and the Yoruba populations (see Supplementary file 3 - Table S3). Overall, the simulated BGS patterns qualitatively match the observation very well (Figure 4, Figure 4—figure supplement 1, and Figure 4—figure supplement 2). As observed in real data (Figure 1), neutral sites simulated next to selected regions present a strong increase in ${\displaystyle \overline{DA{F}_i}}$ with recombination rate (Figure 4A), and the SFS at neutral sites shows a considerable excess of singletons and a deficit of intermediate- and high-frequency variants for low-recombination rates (Figure 4B), respectively. These results show that BGS can reproduce both the observed correlation between ${\displaystyle \overline{DA{F}_i}}$ and local recombination rates, and the observed distortions of the SFS in low-recombination regions.

Figure 4 with 2 supplements see all

Download asset Open asset

Delineating the neutrally evolving part of the human genome remains a challenge, as variation in the intensity of recombination, mutation, and selection are increasingly recognised as having a strong effect on observable genomic diversity in humans (Corbett-Detig et al., 2015; Elyashiv et al., 2016) and other organisms (e.g. Elyashiv et al., 2016; Ravinet et al., 2017). Here, we have shown that a surprisingly large proportion (up to 95%) of our genome might be affected by background selection (BGS) and/or GC-biased gene conversion (gBGC). These two processes, which both depend on recombination, strongly affect observed measures of genetic diversity along the genome and can lead to biased demographic inference if not properly taken into account (Figure 3).

We have interpreted the striking linear relationship observed between ${\displaystyle \overline{DA{F}_i}}$ and recombination rate (Figure 1) as evidence for the pervasive effect of BGS but other processes could in principle lead to a similar relationship. For instance, a mutagenic effect of recombination could lead to an increased diversity in regions of high recombination (Hellmann et al., 2003). The examination of extremely low-frequency mutations, which should be enriched for new mutations, did not reveal any association between recombination rate and the density of new mutations in a large human sample (Schaibley et al., 2013), but a more recent study of de novo mutations suggested the existence of such a correlation (Francioli et al., 2015). Alternatively, a correlation between mutation and recombination rates could occur if these rates were both affected by the same process, such as replication timing (Stamatoyannopoulos et al., 2009; Koren et al., 2012) or transcription rate (Gerton et al., 2000; Park et al., 2012). However, a mere correlation between mutation and recombination rates cannot explain two key aspects of our observations. First, ${\displaystyle \overline{DA{F}_i}}$ plateau at high recombination rates once the effect of gBGC is removed (Figure 1C), whereas it should continue increasing if only mutation-recombination correlation was driving the relationship between ${\displaystyle \overline{DA{F}_i}}$ and recombination. Second, we find a significant difference in the shape of SFS computed in regions of low and high recombination (Figure 2A), even though mutation rate should have no effect on the shape of the SFS. To better investigate the effect of a possible mutation-recombination correlation, we have used the fact that ${\displaystyle \overline{DA{F}_i}}$ is correlated with the B-statistic (Figure 1—figure supplement 5F), for which a simple model (Hudson and Kaplan, 1995) predicts its value as a function of mutation and recombination rates. We find that the B-statistics inferred by McVicker et al. (2009) are significantly better fitted as a function of the recombination rate if we assume a log-log linear relationship between recombination and deleterious mutation rate than if we impose a constant mutation rate across the genome (Figure 1—figure supplement 8). Interestingly, under our log-log linear model, both the observed and predicted B-statistics reach a plateau value of ~0.9 above a recombination rate of ~1.5 cM/Mb. This pattern remains if we only consider subsets of SNPs (e.g. WW + SS sites; Supplementary file 4 - Table S4). Therefore, these results suggest that in addition to BGS and gBGC, some correlation between mutation and recombination rate is required to best explain our observed patterns. Moreover, given the relationship observed between B-statistics and ${\displaystyle \overline{DA{F}_i}}$ (Figure 1—figure supplement 5F), the reduced effect of recombination on B above 1.5 cM/Mb should translate into a similar absence of change in ${\displaystyle \overline{DA{F}_i}}$ above the same threshold, thus explaining the plateau we see in Figure 1C above 1.5 cM/Mb.

The occurrence of pervasive positive selection, either in the form of soft or hard sweeps (Kern and Hahn, 2018) or of positive selection on polygenic traits (Boyle et al., 2017) in our genome could also lead to a correlation between genetic diversity and recombination, as the effect of selection on linked neutral sites should decrease with recombination. However, positive selection should lead to an increase of both low- and high-frequency variants in the SFS (Fay et al., 2000; Hernandez et al., 2007; Huber et al., 2016; Pavlidis and Alachiotis, 2017), whereas we only observe an increase of low-frequency variants in low-recombination regions where the effect of selection should be strongest (Figure 2A), which is the expected effect of BGS (Figure 4).

The exact proportion of the genome that is influenced by selection is still the source of an intense debate (Bernstein et al., 2012; Rands et al., 2014; Graur, 2017; Kern and Hahn, 2018). Here, we show that up to 80–85% of the human genome is probably affected by background selection (BGS), an effect that is not subtle (Reed et al., 2005) and that is visible from single individuals genomes (Figure 1—figure supplement 5A). Even though our estimate of the fraction of the human genome influenced by BGS matches relatively well with that reported to be biochemically functional by the ENCODE consortium (Bernstein et al., 2012), our results do not imply that 80–85% of the human genome is functional. They rather show that functional sites that are the direct target of purifying selection in both coding and non-coding regions (potentially representing 8–15% of the genome, Rands et al. (2014); Graur, 2017) have an important but indirect influence on most of the genome.

As expected, the effect of BGS is clearly mediated by local recombination rate, but it extends well beyond coding regions in humans (Hernandez et al., 2011) (Figure 1), and it is thus not restricted to species with a large effective size (Corbett-Detig et al., 2015). Our results also show that the influence of gBGC is not restricted to recombination hotspots (Spencer et al., 2006; Glémin et al., 2015), but that it has also a strong footprint in regions with a recombination rate larger than 1.5 cM/Mb, but note that it could affect (to a lesser degree) regions with an even lower recombination rate (see Figure 1—figure supplement 5D. These regions represent about 15.9% and 16.2% of the polymorphic positions for the 1000G and SGDP datasets, respectively. Taken together, BGS and gBGC thus affect more than 95% of the polymorphic sites in our genome, and we have identified only a small fraction of all genomic SNPs (~3%, Supplementary file 2 - Table S2) that can be considered as evolving neutrally.

Interestingly, our neutral SNPs are found in both transcribed and non-transcribed-regions (Figure 2C), and they are enriched close to telomeric regions (Figure 1—figure supplement 10), where BGS is predicted to be weaker (Charlesworth, 2012). Whereas SNPs included in our best-filtered set are evolving mostly neutrally, it does not imply that all other SNPs are influenced by BGS and gBGC. Indeed, our way of identifying selection and biased gene conversion is indirect and operates on arbitrarily defined recombination-rate categories. Thus, ${\displaystyle \overline{DA{F}_i}}$ cannot be used to identify the presence of selection at the SNP level or in small genomic regions, or inversely, the presence of neutral SNPs in low recombining segments between recombination hotspots. A more precise mapping of selected genomic segments could use information on the positions of known functional elements (Siepel et al., 2005; Kellis et al., 2014; Rands et al., 2014; Elkon and Agami, 2017) or B-statistics (McVicker et al., 2009; Elyashiv et al., 2016), which could also be used to evidence neutrally evolving regions in both low- and high-recombination regions.

To investigate if and how ${\displaystyle \overline{DA{F}_i}}$ depends on potential co-variates within our neutral set of SNPs, we have examined its relationship with several statistics, such as B-statistics or the distance (in map units) to exons, as well as distances to conserved elements and to recombination hotspots. In our neutral set, we find virtually no relationship between ${\displaystyle \overline{DA{F}_i}}$ and recombination rate, with average ${\displaystyle \overline{DA{F}_i}}$ remaining close to its mean value of 0.146 (Figure 1—figure supplement 9A), but we find a negative relation with the distance to recombination hotspots (Figure 1—figure supplement 9B, a positive relationship with distance to conserved elements and with B-statistics (Figure 1—figure supplement 9C–D), and a small positive correlation with distance to exons (${\displaystyle \overline{DA{F}_i}}$ varies from 0.145 to 0.15, close to the average, Figure 1—figure supplement 9E). It thus seems that recombination hotspots still play a role in decoupling selected from neutral sites, and that sites furthest away from hotspots might still be slightly sensitive to BGS. Purifying selection in phastCons conserved elements (Siepel et al., 2005) is also exerting a strong negative pressure on derived allele frequencies, with average ${\displaystyle \overline{DA{F}_i}}$ below 0.14 at sites less than 0.0003 cM away from these elements (which correponds approximatively to a distance of 200 bp if RR = 1.5 cM/Mb). Contrastingly, being further than 0.05 cM away from these conserved elements allows ${\displaystyle \overline{DA{F}_i}}$ to rise above 0.16, an average value that is barely reached for sites with associated mean B values close to 1. These results suggest that phastCons elements represent the covariate that has the strongest remaining influence on ${\displaystyle \overline{DA{F}_i}}$ within our neutral set.

The SFS of each population is affected by BGS and gBGC (Figure 2, Figure 2—figure supplement 1), and the demography inferred from neutrally evolving SNPs differs markedly from that based on synonymous sites or sites in non-transcribed regions (Figure 3A). However, we show that BGS and gBGC can have different impacts on the inferred demography of the populations. For instance, we found that they lead to an underestimation of the age of a bottleneck and an overestimation of the magnitude of a demographic expansion in the Yoruba population, but we do not observe such strong biases in the Japanese population. It therefore appears difficult to predict the specific biases introduced by these evolutionary forces on demographic inference, except perhaps under simple evolutionary scenarios (Ewing and Jensen, 2016). We therefore suggest that future studies of demographic history should be based on a set of markers that is minimally influenced by these non-neutral forces.

We have also computed the observed SFS for subsets of neutral SNPs with various values of the covariates mentioned above (Figure 1—figure supplement 9). SNPs in the 1st and 4th distance-quartiles to hotspot show similar SFS, with a slight excess of singletons and high-frequency variants for the sites furthest to hotspots (Figure 2—figure supplement 3A. Even though conserved elements had the strongest influence on ${\displaystyle \overline{DA{F}_i}}$, the SFSs computed at sites belonging to the 4th distance quartile and to all sites still look very similar, especially in the Japanese population, while sites in the 1st distance quartileshow an excess of singletons and a deficit of high-frequency variants (Figure 2—figure supplement 3B. Exonic and non-exonic SFSs within our neutral set differ mainly by increased frequencies of singletons for exonic SNPs, yet the removal of exonic SNPs has no impact on the SFS (Figure 2—figure supplement 3C). In conclusion, even though exonic SNPs and those located close (≤0.0003 cM) to phastCons elements show different SFS shapes (Figure 2—figure supplement 4), their removal from our neutral set would have no major effect on the shape of the SFS, since they represent only a small fraction (2.2% and 16.9% respectively) of the SNPs in our neutral set.

It is interesting to compare our neutral set of SNPs to another previously defined set of neutral regions of the human genome that has been used as a reference for demographic inferences in a series of studies (e.g. Gronau et al., 2011; McManus et al., 2015; King and Wakeley, 2016; Veeramah et al., 2018). Gronau et al. (2011) have identified a set of 37,574 potentially neutral regions of 1 kb in length with carefully chosen properties (e.g. at least 1 kb away from exons and 100 bp away from phastCons elements, without CpG sites, separated by at least 50 kb, without recombination hotspots). The SFS computed on this alternative neutral set departs significantly from our neutral set, with a significant excess of singletons, a deficit of sites with intermediate allele frequencies, and an excess of nearly fixed variants, a pattern that can be explained by the action of both BGS and gBGC (Figure 3—figure supplement 3A. Since a large B-statistic is also indicative of relaxed BGS, one could be tempted to use regions associated with B values larger than 0.9 as being potentially neutral. However, we see that its SFS also departs from that of our neutral set, with a small deficit of singleton and an excess of other frequency classes in Yoruba, and a slight excess of high-frequency variants in Japan (Figure 3—figure supplement 3A). These differences in SFS shapes also lead to inferred demographies that are markedly different from that inferred from our own neutral set, and this especially for the Yoruba population (Figure 3—figure supplement 3B). We suspect that the main discrepancy with our neutral set is the presence of gBGC in regions with B > 0.9, such that filtering out SW and WS SNPs may result in a good alternative data set on which to perform demographic inferences

Methods of demographic inference based on whole genomes (e.g. Li and Durbin, 2009; Sheehan et al., 2013; Schiffels and Durbin, 2014) should also be sensitive to BGS and gBGC, since they assume that heterozygosity levels within individuals is not driven by local recombination rates nor selection. In this respect, the history of human populations as well as that of other species might be more readily inferred from methods that can conveniently analyze restricted sets of neutrally evolving sites interspersed across the genome. Similarly, other types of inference using a biased neutral SFS as a reference could also be affected, such as inferences of the distributions of fitness effects (DFE) (Keightley and Eyre-Walker, 2010; Kim et al., 2017; Tataru et al., 2017), even though the magnitude of the effect remains to be investigated. In conclusion, we show that BGS and gBGC had a pervasive effect on most of our genome, but that we can conveniently define a set of sites (representing about 3% of all polymorphic sites of both 1000G and SGDP datasets) that should not be too influenced by these two evolutionary forces, even though some sites close to conserved elements could still be affected by BGS. Contrary to previously used sets of SNPs, these sites should lead to essentially unbiased demographic inferences and serve as a reference for future demographic reconstructions in humans. Due to its simplicity, our approach can be readily applied to any species for which a recombination map is available.

All data generated and script to analyse them is provided on the dryad repository: http://dx.doi.org/10.5061/dryad.t76fk80.

1. 1. Fanny Pouyet
   2. Simon Aeschbacher
   3. Alexandre Thiéry
   4. Laurent Excoffier

   (2018) Data from: Background selection and biased gene conversion affect more than 95% of the human genome and bias demographic inferences

   Available at Dryad Digital Repository under a CC0 Public Domain Dedication.

   http://dx.doi.org/10.5061/dryad.t76fk80

1. 1. Mallick S
   2. Li H
   3. Lipson M
   4. Mathieson I
   5. Gimrek M
   6. et al.

   (2016) Simons Genome Diversity Panel

   Publicly available at Cancer Genomics Cloud (https://www.cancergenomicscloud.org/).

   https://docs.cancergenomicscloud.org/v1.0/docs/simons-genome-diversity-project-sgdp-dataset
2. 1. Auton A
   2. Brooks RD
   3. Durbin RM
   4. Garrison EP
   5. Kang HM
   6. et al.

   (2015) 1000G Project

   Publicly available at The International Genome Sample Resource (http://internationalgenome.org/home).

   ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/

1. 1. Akaike H

   (1974) A new look at the statistical model identification

   IEEE Transactions on Automatic Control 19:716–723.

   https://doi.org/10.1109/TAC.1974.1100705

   • Google Scholar
2. 1. Arbeithuber B
   2. Betancourt AJ
   3. Ebner T
   4. Tiemann-Boege I

   (2015) Crossovers are associated with mutation and biased gene conversion at recombination hotspots

   PNAS 112:2109–2114.

   https://doi.org/10.1073/pnas.1416622112

   • PubMed
   • Google Scholar
3. 1. Asthana S
   2. Noble WS
   3. Kryukov G
   4. Grant CE
   5. Sunyaev S
   6. Stamatoyannopoulos JA

   (2007) Widely distributed noncoding purifying selection in the human genome

   PNAS 104:12410–12415.

   https://doi.org/10.1073/pnas.0705140104

   • PubMed
   • Google Scholar
4. 1. Auton A
   2. Brooks LD
   3. Durbin RM
   4. Garrison EP
   5. Kang HM
   6. Korbel JO
   7. Marchini JL
   8. McCarthy S
   9. McVean GA
   10. Abecasis GR
   11. 1000 Genomes Project Consortium

   (2015) A global reference for human genetic variation

   Nature 526:68–74.

   https://doi.org/10.1038/nature15393

   • PubMed
   • Google Scholar
5. 1. Bernstein BE
   2. Birney E
   3. Dunham I
   4. Green ED
   5. Gunter C
   6. Snyder M
   7. ENCODE Project Consortium

   (2012) An integrated encyclopedia of DNA elements in the human genome

   Nature 489:57–74.

   https://doi.org/10.1038/nature11247

   • PubMed
   • Google Scholar
6. 1. Bonev B
   2. Cavalli G

   (2016) Organization and function of the 3D genome

   Nature Reviews Genetics 17:661–678.

   https://doi.org/10.1038/nrg.2016.112

   • PubMed
   • Google Scholar
7. 1. Boyle EA
   2. Li YI
   3. Pritchard JK

   (2017) An expanded view of complex traits: from polygenic to omnigenic

   Cell 169:1177–1186.

   https://doi.org/10.1016/j.cell.2017.05.038

   • PubMed
   • Google Scholar
8. 1. Cai JJ
   2. Macpherson JM
   3. Sella G
   4. Petrov DA

   (2009) Pervasive hitchhiking at coding and regulatory sites in humans

   PLoS Genetics 5:e1000336.

   https://doi.org/10.1371/journal.pgen.1000336

   • PubMed
   • Google Scholar
9. 1. Charlesworth B

   (1994) The effect of background selection against deleterious mutations on weakly selected, linked variants

   Genetical Research 63:213–227.

   https://doi.org/10.1017/S0016672300032365

   • PubMed
   • Google Scholar
10. 1. Charlesworth D
    2. Charlesworth B
    3. Morgan MT

    (1995)

    The pattern of neutral molecular variation under the background selection model

    Genetics 141:1619–1632.

    • PubMed
    • Google Scholar
11. 1. Charlesworth B

    (2012) The effects of deleterious mutations on evolution at linked sites

    Genetics 190:5–22.

    https://doi.org/10.1534/genetics.111.134288

    • PubMed
    • Google Scholar
12. 1. Comeron JM

    (2014) Background selection as baseline for nucleotide variation across the Drosophila genome

    PLoS Genetics 10:e1004434.

    https://doi.org/10.1371/journal.pgen.1004434

    • PubMed
    • Google Scholar
13. 1. Corbett-Detig RB
    2. Hartl DL
    3. Sackton TB

    (2015) Natural selection constrains neutral diversity across a wide range of species

    PLOS Biology 13:e1002112.

    https://doi.org/10.1371/journal.pbio.1002112

    • PubMed
    • Google Scholar
14. 1. Davydov EV
    2. Goode DL
    3. Sirota M
    4. Cooper GM
    5. Sidow A
    6. Batzoglou S

    (2010) Identifying a high fraction of the human genome to be under selective constraint using GERP++

    PLoS Computational Biology 6:e1001025.

    https://doi.org/10.1371/journal.pcbi.1001025

    • PubMed
    • Google Scholar
15. 1. Duret L
    2. Galtier N

    (2009) Biased gene conversion and the evolution of mammalian genomic landscapes

    Annual Review of Genomics and Human Genetics 10:285–311.

    https://doi.org/10.1146/annurev-genom-082908-150001

    • PubMed
    • Google Scholar
16. 1. Elkon R
    2. Agami R

    (2017) Characterization of noncoding regulatory DNA in the human genome

    Nature Biotechnology 35:732–746.

    https://doi.org/10.1038/nbt.3863

    • PubMed
    • Google Scholar
17. 1. Elyashiv E
    2. Sattath S
    3. Hu TT
    4. Strutsovsky A
    5. McVicker G
    6. Andolfatto P
    7. Coop G
    8. Sella G

    (2016) A genomic map of the effects of linked selection in Drosophila

    PLOS Genetics 12:e1006130.

    https://doi.org/10.1371/journal.pgen.1006130

    • PubMed
    • Google Scholar
18. 1. Ewing GB
    2. Jensen JD

    (2016) The consequences of not accounting for background selection in demographic inference

    Molecular Ecology 25:135–141.

    https://doi.org/10.1111/mec.13390

    • PubMed
    • Google Scholar
19. 1. Excoffier L
    2. Dupanloup I
    3. Huerta-Sánchez E
    4. Sousa VC
    5. Foll M

    (2013) Robust demographic inference from genomic and SNP data

    PLoS Genetics 9:e1003905.

    https://doi.org/10.1371/journal.pgen.1003905

    • PubMed
    • Google Scholar
20. 1. Fay JC
    2. Wu CI
    3. Ci W

    (2000)

    Hitchhiking under positive Darwinian selection

    Genetics 155:1405–1413.

    • PubMed
    • Google Scholar
21. 1. Francioli LC
    2. Polak PP
    3. Koren A
    4. Menelaou A
    5. Chun S
    6. Renkens I
    7. van Duijn CM
    8. Swertz M
    9. Wijmenga C
    10. van Ommen G
    11. Slagboom PE
    12. Boomsma DI
    13. Ye K
    14. Guryev V
    15. Arndt PF
    16. Kloosterman WP
    17. de Bakker PIW
    18. Sunyaev SR
    19. Genome of the Netherlands Consortium

    (2015) Genome-wide patterns and properties of de novo mutations in humans

    Nature Genetics 47:822–826.

    https://doi.org/10.1038/ng.3292

    • PubMed
    • Google Scholar
22. 1. Frazer KA
    2. Ballinger DG
    3. Cox DR
    4. Hinds DA
    5. Stuve LL
    6. Gibbs RA
    7. Belmont JW
    8. Boudreau A
    9. Hardenbol P
    10. Leal SM
    11. Pasternak S
    12. Wheeler DA
    13. Willis TD
    14. Yu F
    15. Yang H
    16. Zeng C
    17. Gao Y
    18. Hu H
    19. Hu W
    20. Li C
    21. Lin W
    22. Liu S
    23. Pan H
    24. Tang X
    25. Wang J
    26. Wang W
    27. Yu J
    28. Zhang B
    29. Zhang Q
    30. Zhao H
    31. Zhao H
    32. Zhou J
    33. Gabriel SB
    34. Barry R
    35. Blumenstiel B
    36. Camargo A
    37. Defelice M
    38. Faggart M
    39. Goyette M
    40. Gupta S
    41. Moore J
    42. Nguyen H
    43. Onofrio RC
    44. Parkin M
    45. Roy J
    46. Stahl E
    47. Winchester E
    48. Ziaugra L
    49. Altshuler D
    50. Shen Y
    51. Yao Z
    52. Huang W
    53. Chu X
    54. He Y
    55. Jin L
    56. Liu Y
    57. Shen Y
    58. Sun W
    59. Wang H
    60. Wang Y
    61. Wang Y
    62. Xiong X
    63. Xu L
    64. Waye MM
    65. Tsui SK
    66. Xue H
    67. Wong JT
    68. Galver LM
    69. Fan JB
    70. Gunderson K
    71. Murray SS
    72. Oliphant AR
    73. Chee MS
    74. Montpetit A
    75. Chagnon F
    76. Ferretti V
    77. Leboeuf M
    78. Olivier JF
    79. Phillips MS
    80. Roumy S
    81. Sallée C
    82. Verner A
    83. Hudson TJ
    84. Kwok PY
    85. Cai D
    86. Koboldt DC
    87. Miller RD
    88. Pawlikowska L
    89. Taillon-Miller P
    90. Xiao M
    91. Tsui LC
    92. Mak W
    93. Song YQ
    94. Tam PK
    95. Nakamura Y
    96. Kawaguchi T
    97. Kitamoto T
    98. Morizono T
    99. Nagashima A
    100. Ohnishi Y
    101. Sekine A
    102. Tanaka T
    103. Tsunoda T
    104. Deloukas P
    105. Bird CP
    106. Delgado M
    107. Dermitzakis ET
    108. Gwilliam R
    109. Hunt S
    110. Morrison J
    111. Powell D
    112. Stranger BE
    113. Whittaker P
    114. Bentley DR
    115. Daly MJ
    116. de Bakker PI
    117. Barrett J
    118. Chretien YR
    119. Maller J
    120. McCarroll S
    121. Patterson N
    122. Pe'er I
    123. Price A
    124. Purcell S
    125. Richter DJ
    126. Sabeti P
    127. Saxena R
    128. Schaffner SF
    129. Sham PC
    130. Varilly P
    131. Altshuler D
    132. Stein LD
    133. Krishnan L
    134. Smith AV
    135. Tello-Ruiz MK
    136. Thorisson GA
    137. Chakravarti A
    138. Chen PE
    139. Cutler DJ
    140. Kashuk CS
    141. Lin S
    142. Abecasis GR
    143. Guan W
    144. Li Y
    145. Munro HM
    146. Qin ZS
    147. Thomas DJ
    148. McVean G
    149. Auton A
    150. Bottolo L
    151. Cardin N
    152. Eyheramendy S
    153. Freeman C
    154. Marchini J
    155. Myers S
    156. Spencer C
    157. Stephens M
    158. Donnelly P
    159. Cardon LR
    160. Clarke G
    161. Evans DM
    162. Morris AP
    163. Weir BS
    164. Tsunoda T
    165. Mullikin JC
    166. Sherry ST
    167. Feolo M
    168. Skol A
    169. Zhang H
    170. Zeng C
    171. Zhao H
    172. Matsuda I
    173. Fukushima Y
    174. Macer DR
    175. Suda E
    176. Rotimi CN
    177. Adebamowo CA
    178. Ajayi I
    179. Aniagwu T
    180. Marshall PA
    181. Nkwodimmah C
    182. Royal CD
    183. Leppert MF
    184. Dixon M
    185. Peiffer A
    186. Qiu R
    187. Kent A
    188. Kato K
    189. Niikawa N
    190. Adewole IF
    191. Knoppers BM
    192. Foster MW
    193. Clayton EW
    194. Watkin J
    195. Gibbs RA
    196. Belmont JW
    197. Muzny D
    198. Nazareth L
    199. Sodergren E
    200. Weinstock GM
    201. Wheeler DA
    202. Yakub I
    203. Gabriel SB
    204. Onofrio RC
    205. Richter DJ
    206. Ziaugra L
    207. Birren BW
    208. Daly MJ
    209. Altshuler D
    210. Wilson RK
    211. Fulton LL
    212. Rogers J
    213. Burton J
    214. Carter NP
    215. Clee CM
    216. Griffiths M
    217. Jones MC
    218. McLay K
    219. Plumb RW
    220. Ross MT
    221. Sims SK
    222. Willey DL
    223. Chen Z
    224. Han H
    225. Kang L
    226. Godbout M
    227. Wallenburg JC
    228. L'Archevêque P
    229. Bellemare G
    230. Saeki K
    231. Wang H
    232. An D
    233. Fu H
    234. Li Q
    235. Wang Z
    236. Wang R
    237. Holden AL
    238. Brooks LD
    239. McEwen JE
    240. Guyer MS
    241. Wang VO
    242. Peterson JL
    243. Shi M
    244. Spiegel J
    245. Sung LM
    246. Zacharia LF
    247. Collins FS
    248. Kennedy K
    249. Jamieson R
    250. Stewart J
    251. International HapMap Consortium

    (2007) A second generation human haplotype map of over 3.1 million SNPs

    Nature 449:851–861.

    https://doi.org/10.1038/nature06258

    • PubMed
    • Google Scholar
23. 1. Fu W
    2. O'Connor TD
    3. Jun G
    4. Kang HM
    5. Abecasis G
    6. Leal SM
    7. Gabriel S
    8. Rieder MJ
    9. Altshuler D
    10. Shendure J
    11. Nickerson DA
    12. Bamshad MJ
    13. Akey JM
    14. NHLBI Exome Sequencing Project

    (2013) Analysis of 6,515 exomes reveals the recent origin of most human protein-coding variants

    Nature 493:216–220.

    https://doi.org/10.1038/nature11690

    • PubMed
    • Google Scholar
24. 1. Galtier N
    2. Duret L

    (2007) Adaptation or biased gene conversion? Extending the null hypothesis of molecular evolution

    Trends in Genetics 23:273–277.

    https://doi.org/10.1016/j.tig.2007.03.011

    • PubMed
    • Google Scholar
25. 1. Gerton JL
    2. DeRisi J
    3. Shroff R
    4. Lichten M
    5. Brown PO
    6. Petes TD

    (2000) Global mapping of meiotic recombination hotspots and coldspots in the yeast Saccharomyces cerevisiae

    PNAS 97:11383–11390.

    https://doi.org/10.1073/pnas.97.21.11383

    • PubMed
    • Google Scholar
26. 1. Glémin S
    2. Arndt PF
    3. Messer PW
    4. Petrov D
    5. Galtier N
    6. Duret L

    (2015) Quantification of GC-biased gene conversion in the human genome

    Genome Research 25:1215–1228.

    https://doi.org/10.1101/gr.185488.114

    • PubMed
    • Google Scholar
27. 1. Graur D

    (2017) An upper limit on the functional fraction of the human genome

    Genome Biology and Evolution 9:1880–1885.

    https://doi.org/10.1093/gbe/evx121

    • PubMed
    • Google Scholar
28. 1. Gronau I
    2. Hubisz MJ
    3. Gulko B
    4. Danko CG
    5. Siepel A

    (2011) Bayesian inference of ancient human demography from individual genome sequences

    Nature Genetics 43:1031–1034.

    https://doi.org/10.1038/ng.937

    • PubMed
    • Google Scholar
29. 1. Haller BC
    2. Messer PW

    (2017) SLiM 2: flexible, interactive forward genetic simulations

    Molecular Biology and Evolution 34:230–240.

    https://doi.org/10.1093/molbev/msw211

    • PubMed
    • Google Scholar
30. 1. Hellmann I
    2. Ebersberger I
    3. Ptak SE
    4. Pääbo S
    5. Przeworski M

    (2003) A neutral explanation for the correlation of diversity with recombination rates in humans

    The American Journal of Human Genetics 72:1527–1535.

    https://doi.org/10.1086/375657

    • PubMed
    • Google Scholar
31. 1. Hernandez RD
    2. Williamson SH
    3. Zhu L
    4. Bustamante CD

    (2007) Context-dependent mutation rates may cause spurious signatures of a fixation bias favoring higher GC-content in humans

    Molecular Biology and Evolution 24:2196–2202.

    https://doi.org/10.1093/molbev/msm149

    • PubMed
    • Google Scholar
32. 1. Hernandez RD
    2. Kelley JL
    3. Elyashiv E
    4. Melton SC
    5. Auton A
    6. McVean G
    7. Sella G
    8. Przeworski M
    9. 1000 Genomes Project

    (2011) Classic selective sweeps were rare in recent human evolution

    Science 331:920–924.

    https://doi.org/10.1126/science.1198878

    • PubMed
    • Google Scholar
33. 1. Huber CD
    2. DeGiorgio M
    3. Hellmann I
    4. Nielsen R

    (2016) Detecting recent selective sweeps while controlling for mutation rate and background selection

    Molecular Ecology 25:142–156.

    https://doi.org/10.1111/mec.13351

    • PubMed
    • Google Scholar
34. 1. Hudson RR
    2. Kaplan NL

    (1995)

    Deleterious background selection with recombination

    Genetics 141:1605–1617.

    • PubMed
    • Google Scholar
35. 1. Katzman S
    2. Capra JA
    3. Haussler D
    4. Pollard KS

    (2011) Ongoing GC-biased evolution is widespread in the human genome and enriched near recombination hot spots

    Genome Biology and Evolution 3:614–626.

    https://doi.org/10.1093/gbe/evr058

    • PubMed
    • Google Scholar
36. 1. Keightley PD
    2. Eyre-Walker A

    (2010) What can we learn about the distribution of fitness effects of new mutations from DNA sequence data?

    Philosophical Transactions of the Royal Society B: Biological Sciences 365:1187–1193.

    https://doi.org/10.1098/rstb.2009.0266

    • PubMed
    • Google Scholar
37. 1. Kellis M
    2. Wold B
    3. Snyder MP
    4. Bernstein BE
    5. Kundaje A
    6. Marinov GK
    7. Ward LD
    8. Birney E
    9. Crawford GE
    10. Dekker J
    11. Dunham I
    12. Elnitski LL
    13. Farnham PJ
    14. Feingold EA
    15. Gerstein M
    16. Giddings MC
    17. Gilbert DM
    18. Gingeras TR
    19. Green ED
    20. Guigo R
    21. Hubbard T
    22. Kent J
    23. Lieb JD
    24. Myers RM
    25. Pazin MJ
    26. Ren B
    27. Stamatoyannopoulos JA
    28. Weng Z
    29. White KP
    30. Hardison RC

    (2014) Defining functional DNA elements in the human genome

    PNAS 111:6131–6138.

    https://doi.org/10.1073/pnas.1318948111

    • PubMed
    • Google Scholar
38. 1. Kern AD
    2. Hahn MW

    (2018) The neutral theory in light of natural selection

    Molecular Biology and Evolution 35:1366–1371.

    https://doi.org/10.1093/molbev/msy092

    • PubMed
    • Google Scholar
39. 1. Kim BY
    2. Huber CD
    3. Lohmueller KE

    (2017) Inference of the Distribution of Selection Coefficients for New Nonsynonymous Mutations Using Large Samples

    Genetics 206:345–361.

    https://doi.org/10.1534/genetics.116.197145

    • PubMed
    • Google Scholar
40. 1. King L
    2. Wakeley J

    (2016) Empirical bayes estimation of coalescence times from nucleotide sequence data

    Genetics 204:249–257.

    https://doi.org/10.1534/genetics.115.185751

    • PubMed
    • Google Scholar
41. 1. Kong A
    2. Thorleifsson G
    3. Gudbjartsson DF
    4. Masson G
    5. Sigurdsson A
    6. Jonasdottir A
    7. Walters GB
    8. Jonasdottir A
    9. Gylfason A
    10. Kristinsson KT
    11. Gudjonsson SA
    12. Frigge ML
    13. Helgason A
    14. Thorsteinsdottir U
    15. Stefansson K

    (2010) Fine-scale recombination rate differences between sexes, populations and individuals

    Nature 467:1099–1103.

    https://doi.org/10.1038/nature09525

    • PubMed
    • Google Scholar
42. 1. Koren A
    2. Polak P
    3. Nemesh J
    4. Michaelson JJ
    5. Sebat J
    6. Sunyaev SR
    7. McCarroll SA

    (2012) Differential relationship of DNA replication timing to different forms of human mutation and variation

    The American Journal of Human Genetics 91:1033–1040.

    https://doi.org/10.1016/j.ajhg.2012.10.018

    • PubMed
    • Google Scholar
43. 1. Lachance J
    2. Tishkoff SA

    (2014) Biased gene conversion skews allele frequencies in human populations, increasing the disease burden of recessive alleles

    The American Journal of Human Genetics 95:408–420.

    https://doi.org/10.1016/j.ajhg.2014.09.008

    • PubMed
    • Google Scholar
44. 1. Lapierre M
    2. Lambert A
    3. Achaz G

    (2017) Accuracy of demographic inferences from the site frequency spectrum: the case of the yoruba population

    Genetics 206:439–449.

    https://doi.org/10.1534/genetics.116.192708

    • PubMed
    • Google Scholar
45. 1. Li H
    2. Durbin R

    (2009) Fast and accurate short read alignment with Burrows-Wheeler transform

    Bioinformatics 25:1754–1760.

    https://doi.org/10.1093/bioinformatics/btp324

    • PubMed
    • Google Scholar
46. 1. Liu Q
    2. Zhou Y
    3. Morrell PL
    4. Gaut BS

    (2017) Deleterious variants in asian rice and the potential cost of domestication

    Molecular Biology and Evolution 34:908–924.

    https://doi.org/10.1093/molbev/msw296

    • PubMed
    • Google Scholar
47. 1. Lohmueller KE
    2. Albrechtsen A
    3. Li Y
    4. Kim SY
    5. Korneliussen T
    6. Vinckenbosch N
    7. Tian G
    8. Huerta-Sanchez E
    9. Feder AF
    10. Grarup N
    11. Jørgensen T
    12. Jiang T
    13. Witte DR
    14. Sandbæk A
    15. Hellmann I
    16. Lauritzen T
    17. Hansen T
    18. Pedersen O
    19. Wang J
    20. Nielsen R

    (2011) Natural selection affects multiple aspects of genetic variation at putatively neutral sites across the human genome

    PLoS Genetics 7:e1002326.

    https://doi.org/10.1371/journal.pgen.1002326

    • PubMed
    • Google Scholar
48. 1. Mallick S
    2. Li H
    3. Lipson M
    4. Mathieson I
    5. Gymrek M
    6. Racimo F
    7. Zhao M
    8. Chennagiri N
    9. Nordenfelt S
    10. Tandon A
    11. Skoglund P
    12. Lazaridis I
    13. Sankararaman S
    14. Fu Q
    15. Rohland N
    16. Renaud G
    17. Erlich Y
    18. Willems T
    19. Gallo C
    20. Spence JP
    21. Song YS
    22. Poletti G
    23. Balloux F
    24. van Driem G
    25. de Knijff P
    26. Romero IG
    27. Jha AR
    28. Behar DM
    29. Bravi CM
    30. Capelli C
    31. Hervig T
    32. Moreno-Estrada A
    33. Posukh OL
    34. Balanovska E
    35. Balanovsky O
    36. Karachanak-Yankova S
    37. Sahakyan H
    38. Toncheva D
    39. Yepiskoposyan L
    40. Tyler-Smith C
    41. Xue Y
    42. Abdullah MS
    43. Ruiz-Linares A
    44. Beall CM
    45. Di Rienzo A
    46. Jeong C
    47. Starikovskaya EB
    48. Metspalu E
    49. Parik J
    50. Villems R
    51. Henn BM
    52. Hodoglugil U
    53. Mahley R
    54. Sajantila A
    55. Stamatoyannopoulos G
    56. Wee JT
    57. Khusainova R
    58. Khusnutdinova E
    59. Litvinov S
    60. Ayodo G
    61. Comas D
    62. Hammer MF
    63. Kivisild T
    64. Klitz W
    65. Winkler CA
    66. Labuda D
    67. Bamshad M
    68. Jorde LB
    69. Tishkoff SA
    70. Watkins WS
    71. Metspalu M
    72. Dryomov S
    73. Sukernik R
    74. Singh L
    75. Thangaraj K
    76. Pääbo S
    77. Kelso J
    78. Patterson N
    79. Reich D

    (2016) The Simons Genome Diversity Project: 300 genomes from 142 diverse populations

    Nature 538:201–206.

    https://doi.org/10.1038/nature18964

    • PubMed
    • Google Scholar
49. 1. McManus KF
    2. Kelley JL
    3. Song S
    4. Veeramah KR
    5. Woerner AE
    6. Stevison LS
    7. Ryder OA
    8. Ape Genome Project G
    9. Kidd JM
    10. Wall JD
    11. Bustamante CD
    12. Hammer MF

    (2015) Inference of gorilla demographic and selective history from whole-genome sequence data

    Molecular Biology and Evolution 32:600–612.

    https://doi.org/10.1093/molbev/msu394

    • PubMed
    • Google Scholar
50. 1. McVicker G
    2. Gordon D
    3. Davis C
    4. Green P

    (2009) Widespread genomic signatures of natural selection in hominid evolution

    PLoS Genetics 5:e1000471.

    https://doi.org/10.1371/journal.pgen.1000471

    • PubMed
    • Google Scholar
51. 1. Nachman MW

    (2001) Single nucleotide polymorphisms and recombination rate in humans

    Trends in Genetics 17:481–485.

    https://doi.org/10.1016/S0168-9525(01)02409-X

    • PubMed
    • Google Scholar
52. 1. Nicolaisen LE
    2. Desai MM

    (2013) Distortions in genealogies due to purifying selection and recombination

    Genetics 195:221–230.

    https://doi.org/10.1534/genetics.113.152983

    • PubMed
    • Google Scholar
53. 1. Nielsen R
    2. Williamson S
    3. Kim Y
    4. Hubisz MJ
    5. Clark AG
    6. Bustamante C

    (2005) Genomic scans for selective sweeps using SNP data

    Genome Research 15:1566–1575.

    https://doi.org/10.1101/gr.4252305

    • PubMed
    • Google Scholar
54. 1. Nordborg M
    2. Charlesworth B
    3. Charlesworth D

    (1996) The effect of recombination on background selection

    Genetical Research 67:159–174.

    https://doi.org/10.1017/S0016672300033619

    • PubMed
    • Google Scholar
55. 1. Park C
    2. Qian W
    3. Zhang J

    (2012) Genomic evidence for elevated mutation rates in highly expressed genes

    EMBO reports 13:1123–1129.

    https://doi.org/10.1038/embor.2012.165

    • PubMed
    • Google Scholar
56. 1. Pavlidis P
    2. Alachiotis N

    (2017) A survey of methods and tools to detect recent and strong positive selection

    Journal of Biological Research-Thessaloniki 24:.

    https://doi.org/10.1186/s40709-017-0064-0

    • PubMed
    • Google Scholar
57. Software

    1. R core Team

    (2018)

    R: A language and environment for statistical computing

    R Foundation for Statistical Computing, Vienna, Austria.
58. 1. Rands CM
    2. Meader S
    3. Ponting CP
    4. Lunter G

    (2014) 8.2% of the Human genome is constrained: variation in rates of turnover across functional element classes in the human lineage

    PLoS Genetics 10:e1004525.

    https://doi.org/10.1371/journal.pgen.1004525

    • PubMed
    • Google Scholar
59. 1. Ratnakumar A
    2. Mousset S
    3. Glémin S
    4. Berglund J
    5. Galtier N
    6. Duret L
    7. Webster MT

    (2010) Detecting positive selection within genomes: the problem of biased gene conversion

    Philosophical Transactions of the Royal Society B: Biological Sciences 365:2571–2580.

    https://doi.org/10.1098/rstb.2010.0007

    • PubMed
    • Google Scholar
60. 1. Ravinet M
    2. Faria R
    3. Butlin RK
    4. Galindo J
    5. Bierne N
    6. Rafajlović M
    7. Noor MAF
    8. Mehlig B
    9. Westram AM

    (2017) Interpreting the genomic landscape of speciation: a road map for finding barriers to gene flow

    Journal of Evolutionary Biology 30:1450–1477.

    https://doi.org/10.1111/jeb.13047

    • PubMed
    • Google Scholar
61. 1. Reed FA
    2. Akey JM
    3. Aquadro CF

    (2005) Fitting background-selection predictions to levels of nucleotide variation and divergence along the human autosomes

    Genome Research 15:1211–1221.

    https://doi.org/10.1101/gr.3413205

    • PubMed
    • Google Scholar
62. 1. Schaibley VM
    2. Zawistowski M
    3. Wegmann D
    4. Ehm MG
    5. Nelson MR
    6. St Jean PL
    7. Abecasis GR
    8. Novembre J
    9. Zöllner S
    10. Li JZ
    11. Jz L

    (2013) The influence of genomic context on mutation patterns in the human genome inferred from rare variants

    Genome Research 23:1974–1984.

    https://doi.org/10.1101/gr.154971.113

    • PubMed
    • Google Scholar
63. 1. Schiffels S
    2. Durbin R

    (2014) Inferring human population size and separation history from multiple genome sequences

    Nature Genetics 46:919–925.

    https://doi.org/10.1038/ng.3015

    • PubMed
    • Google Scholar
64. 1. Sheehan S
    2. Harris K
    3. Song YS

    (2013) Estimating variable effective population sizes from multiple genomes: a sequentially markov conditional sampling distribution approach

    Genetics 194:647–662.

    https://doi.org/10.1534/genetics.112.149096

    • PubMed
    • Google Scholar
65. 1. Siepel A
    2. Bejerano G
    3. Pedersen JS
    4. Hinrichs AS
    5. Hou M
    6. Rosenbloom K
    7. Clawson H
    8. Spieth J
    9. Hillier LW
    10. Richards S
    11. Weinstock GM
    12. Wilson RK
    13. Gibbs RA
    14. Kent WJ
    15. Miller W
    16. Haussler D

    (2005) Evolutionarily conserved elements in vertebrate, insect, worm, and yeast genomes

    Genome Research 15:1034–1050.

    https://doi.org/10.1101/gr.3715005

    • PubMed
    • Google Scholar
66. 1. Speir ML
    2. Zweig AS
    3. Rosenbloom KR
    4. Raney BJ
    5. Paten B
    6. Nejad P
    7. Lee BT
    8. Learned K
    9. Karolchik D
    10. Hinrichs AS
    11. Heitner S
    12. Harte RA
    13. Haeussler M
    14. Guruvadoo L
    15. Fujita PA
    16. Eisenhart C
    17. Diekhans M
    18. Clawson H
    19. Casper J
    20. Barber GP
    21. Haussler D
    22. Kuhn RM
    23. Kent WJ

    (2016) The UCSC genome browser database: 2016 update

    Nucleic Acids Research 44:D717–D725.

    https://doi.org/10.1093/nar/gkv1275

    • PubMed
    • Google Scholar
67. 1. Spencer CC
    2. Deloukas P
    3. Hunt S
    4. Mullikin J
    5. Myers S
    6. Silverman B
    7. Donnelly P
    8. Bentley D
    9. McVean G

    (2006) The influence of recombination on human genetic diversity

    PLoS Genetics 2:e148.

    https://doi.org/10.1371/journal.pgen.0020148

    • PubMed
    • Google Scholar
68. 1. Stamatoyannopoulos JA
    2. Adzhubei I
    3. Thurman RE
    4. Kryukov GV
    5. Mirkin SM
    6. Sunyaev SR

    (2009) Human mutation rate associated with DNA replication timing

    Nature Genetics 41:393–395.

    https://doi.org/10.1038/ng.363

    • PubMed
    • Google Scholar
69. 1. Stephan W

    (2010) Genetic hitchhiking versus background selection: the controversy and its implications

    Philosophical Transactions of the Royal Society B: Biological Sciences 365:1245–1253.

    https://doi.org/10.1098/rstb.2009.0278

    • PubMed
    • Google Scholar
70. 1. Sudmant PH
    2. Mallick S
    3. Nelson BJ
    4. Hormozdiari F
    5. Krumm N
    6. Huddleston J
    7. Coe BP
    8. Baker C
    9. Nordenfelt S
    10. Bamshad M
    11. Jorde LB
    12. Posukh OL
    13. Sahakyan H
    14. Watkins WS
    15. Yepiskoposyan L
    16. Abdullah MS
    17. Bravi CM
    18. Capelli C
    19. Hervig T
    20. Wee JT
    21. Tyler-Smith C
    22. van Driem G
    23. Romero IG
    24. Jha AR
    25. Karachanak-Yankova S
    26. Toncheva D
    27. Comas D
    28. Henn B
    29. Kivisild T
    30. Ruiz-Linares A
    31. Sajantila A
    32. Metspalu E
    33. Parik J
    34. Villems R
    35. Starikovskaya EB
    36. Ayodo G
    37. Beall CM
    38. Di Rienzo A
    39. Hammer MF
    40. Khusainova R
    41. Khusnutdinova E
    42. Klitz W
    43. Winkler C
    44. Labuda D
    45. Metspalu M
    46. Tishkoff SA
    47. Dryomov S
    48. Sukernik R
    49. Patterson N
    50. Reich D
    51. Eichler EE

    (2015) Global diversity, population stratification, and selection of human copy-number variation

    Science 349:aab3761.

    https://doi.org/10.1126/science.aab3761

    • PubMed
    • Google Scholar
71. 1. Tataru P
    2. Mollion M
    3. Glémin S
    4. Bataillon T

    (2017) Inference of distribution of fitness effects and proportion of adaptive substitutions from polymorphism data

    Genetics 207:1103–1119.

    https://doi.org/10.1534/genetics.117.300323

    • PubMed
    • Google Scholar
72. Preprint

    1. Van Nostrand EL
    2. Freese P
    3. Pratt GA
    4. Wang X
    5. Wei X
    6. Blue SM
    7. Dominguez D
    8. Cody NAL
    9. Olson S
    10. Sundararaman B

    (2017) A Large-Scale binding and functional map of human RNA binding proteins

    bioRxiv.

    https://doi.org/10.1101/179648

    • Google Scholar
73. 1. Veeramah KR
    2. Rott A
    3. Groß M
    4. van Dorp L
    5. López S
    6. Kirsanow K
    7. Sell C
    8. Blöcher J
    9. Wegmann D
    10. Link V
    11. Hofmanová Z
    12. Peters J
    13. Trautmann B
    14. Gairhos A
    15. Haberstroh J
    16. Päffgen B
    17. Hellenthal G
    18. Haas-Gebhard B
    19. Harbeck M
    20. Burger J

    (2018) Population genomic analysis of elongated skulls reveals extensive female-biased immigration in Early Medieval Bavaria

    PNAS 115:3494–3499.

    https://doi.org/10.1073/pnas.1719880115

    • PubMed
    • Google Scholar
74. 1. Wegmann D
    2. Kessner DE
    3. Veeramah KR
    4. Mathias RA
    5. Nicolae DL
    6. Yanek LR
    7. Sun YV
    8. Torgerson DG
    9. Rafaels N
    10. Mosley T
    11. Becker LC
    12. Ruczinski I
    13. Beaty TH
    14. Kardia SL
    15. Meyers DA
    16. Barnes KC
    17. Becker DM
    18. Freimer NB
    19. Novembre J

    (2011) Recombination rates in admixed individuals identified by ancestry-based inference

    Nature Genetics 43:847–853.

    https://doi.org/10.1038/ng.894

    • PubMed
    • Google Scholar
75. 1. Wolfe D
    2. Dudek S
    3. Ritchie MD
    4. Pendergrass SA

    (2013) Visualizing genomic information across chromosomes with PhenoGram

    BioData Mining 6:18.

    https://doi.org/10.1186/1756-0381-6-18

    • PubMed
    • Google Scholar
76. 1. Zeng K
    2. Charlesworth B

    (2011) The joint effects of background selection and genetic recombination on local gene genealogies

    Genetics 189:251–266.

    https://doi.org/10.1534/genetics.111.130575

    • PubMed
    • Google Scholar

Author details

1. Fanny Pouyet

   1. Computational and Molecular Population Genetics, Institute of Ecology and Evolution, University of Bern, Bern, Switzerland
   2. Swiss Institute of Bioinformatics, Lausanne, Switzerland

   Contribution

   Conceptualization, Formal analysis, Investigation, Methodology, Writing—original draft, Writing—review and editing

   Contributed equally with

   Simon Aeschbacher

   For correspondence

   fanny.pouyet@gmail.com

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-5614-6998

2. Simon Aeschbacher

   1. Computational and Molecular Population Genetics, Institute of Ecology and Evolution, University of Bern, Bern, Switzerland
   2. Swiss Institute of Bioinformatics, Lausanne, Switzerland
   3. Department of Evolutionary Biology and Environmental Studies, University of Zurich, Zurich, Switzerland

   Contribution

   Conceptualization, Formal analysis, Investigation, Methodology, Writing—original draft, Writing—review and editing

   Contributed equally with

   Fanny Pouyet

   Competing interests

   No competing interests declared

3. Alexandre Thiéry

   1. Computational and Molecular Population Genetics, Institute of Ecology and Evolution, University of Bern, Bern, Switzerland
   2. Swiss Institute of Bioinformatics, Lausanne, Switzerland

   Contribution

   Data curation

   Competing interests

   No competing interests declared

4. Laurent Excoffier

   1. Computational and Molecular Population Genetics, Institute of Ecology and Evolution, University of Bern, Bern, Switzerland
   2. Swiss Institute of Bioinformatics, Lausanne, Switzerland

   Contribution

   Conceptualization, Resources, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Methodology, Writing—original draft, Project administration, Writing—review and editing

   For correspondence

   laurent.excoffier@iee.unibe.ch

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-7507-6494

• 1,033

  downloads

Views, downloads and citations are aggregated across all versions of this paper published by eLife.