Human Molecular Genetics, 2000, Vol. 9, No. 18 2675-2681
© 2000 Oxford University Press
Integrated analysis of sequence evolution and population history using hypervariable compound haplotypes
Institute of Genetics, University of Nottingham, Queen’s Medical Centre, Nottingham NG7 2UH, UK and 1Department of Genetics, Adrian Building, University of Leicester, University Road, Leicester LE1 7RH, UK
Received 12 July 2000; Revised and Accepted 5 September 2000.
DDBJ/EMBL/GenBank accession nos AJ252012–AJ252014.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTS AND DISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
We have examined compound haplotypes from a highly informative region of human chromosome 16, in which information from the rapid evolution of a highly unstable minisatellite is integrated with data on the longer-term evolution of this segment from 10 flanking substitutional polymorphisms. Combined with sequence data from non-human primates, analysis of relationships between these compound haplotypes allows the reconstruction of a rooted network of the evolutionary pathways between them. Most relationships can be explained via simple substitutional mutations, although the origins of some haplotypes involve recurrent events at a hotspot for substitutional mutation and/or gene conversion. For compound haplotypes including the minisatellite array, the network found in a range of world-wide populations constitutes a highly informative data set for the analysis of population history (437 different compound haplotypes were discriminated among 658 studied). Since the mutation rates and processes of the minisatellite array are known from direct studies, ages for individual lineages have been estimated using associated minisatellite diversity. These analyses suggest that the higher information content and sampling depth of these compound haplotypes may allow more precise calibration of lineage ages than is possible using coalescent analysis of DNA sequence. Using this method we have dated the oldest Eurasian lineage as 52 000–66 000 years and the oldest European specific lineage as 37 600–56 200 years.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTS AND DISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Human genetic diversity is generated in DNA by new mutations, reassorted between chromosomes by recombination and distributed in populations by selection and drift. Studies of modern human chromosomes have generally used patterns of modern diversity to infer population histories, by applying simple models of chromosome evolution. The Y chromosome (1) and mitochondrial DNA (2) present the simplest cases, in which chromosomal evolution can be modelled as the result of sequential mutation to create new haplotypes and recombination can be safely neglected. On the other hand, the absence of recombination from these segments means that observed diversity will not be a simple consequence of mutation and population history, but will inevitably have been distorted by selection.
Recombination may insulate autosomal haplotypes from the effects of strong selection and, if neutrality is assumed, haplotypes of polymorphic loci in regions such as CD4 (3) and DRD2 (4) can be highly informative discriminators of chromosomes with different histories. Similarly, full sequence sampling of chromosomal segments, for example at the ß-globin locus (5) or at the X-linked region studied by Kaessmann et al. (6), allows application of quantitative approaches based on the coalescent to infer population history. Full information on sequence diversity can thus be recovered from a region short enough that recombination will not create insuperable complications. Given that full sequence sampling is laborious, sample size, as well as the information content of a short sequenced region, may be limiting in its application to the study of population history.
This limitation of power may be mitigated by studying short autosomal regions which are nevertheless highly informative (7). MS205 is a hypervariable minisatellite on the short arm of chromosome 16 at which germline mutation occurs at a (sex-average) rate of 
0.4% (8) and is polar in nature: that is, mutations are generally confined to one extremity of the locus, so that closely related alleles will differ only by small changes at this hypermutable 3' end (9–11). Deduction of relatedness between repeat array alleles based on structural comparisons has allowed exploitation of this extremely informative locus in studies of human diversity (12,13). These analyses have, however, been confined to the hypervariable repeat array itself and have not recruited information from the flanking single-copy DNA. In this study we have established compound haplotypes, incorporating information from flanking polymorphisms as well as the hypervariable repeat array, for a large sample of chromosomes from different populations around the world. We have thereby attempted to reconstruct the evolution of this extremely variable segment of human DNA, so that the compound haplotypes resolve differences between lineages both on long and short time-scales. Analysis of relationships between haplotypes suggests that, whereas most changes can be explained by simple substitutional mutations, other relationships require either recurrent substitutional mutation or recombinational mechanisms, such as gene conversion. The extent of repeat array diversity within individual lineages allows ages to be estimated for each lineage. This method of using lineage-restricted diversity to estimate ages and population parameters may have some advantages over more established methods based on coalescent analysis of sequence variation.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTS AND DISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
A total of 389 individuals from different populations (see Materials and Methods) were haplotyped at 10 polymorphic sites shown in Figure 1 using sequential allele-specific oligonucleotide (ASO) hybridization to duplicate dot-blots, followed by establishment of phase in heterozygotes by allele-specific PCR across the minisatellite array. Since the repeat array is generally heterozygous for length, compound haplotypes could be built up by coupling the minisatellite arrays to alleles of flanking substitutional polymorphisms. Considering the 10 flanking polymorphisms alone, 15 haplotypes were observed (Table 1). To root the network showing the relationships between each of these flanking substitutional haplotypes, the ancestral haplotype (Table 2) was deduced from sequence analysis of MS205 5' flanking region in non-human primates (accession nos AJ252012–AJ252014).
|
|
|
Position 8 (8 bp upstream of the 5' end of the minisatellite) is polymorphic (C/T) in chimpanzees as well as in humans, where the polymorphism is also C/T. This chimpanzee polymorphism was confirmed by an NlaIV restriction assay which detects the presence of a C at –8. It seems most likely that this substitution has occurred independently along each branch from the common ancestor—it is unlikely that the polymorphism would have persisted for
5 million years in both species without strong balancing selection. Evidence from analysis of human haplotypes (see below) suggests that this position may indeed be hypermutable.
The network of haplotypes of substitutional polymorphisms was built up initially by joining haplotypes separated by single steps and rooting these paths in the ancestral haplotype A. However, there are some substitutional haplotypes which cannot be derived by single mutational steps and so therefore we must consider recombination or recurrent or revertant substitutions (Figs 2 and 3). Haplotype O can be formed either by a recurrent mutation at –1583 or a recombination between haplotypes N and J. Haplotype P can be formed from haplotype N by a G
A reversion at –1173. For haplotypes A, B, C and D, all four possible combinations of –1173 and –8 genotypes are present, i.e. A–C, G–C, A–T and G–T. This reticulation illustrates that one haplotype must have been formed by a double substitution or recombination event (crossover or gene conversion). In Figure 2 no specific mechanism is singled out, whereas in Figure 3 haplotype C is shown as a recombinant between haplotypes B and D. As haplotype A is the ancestral haplotype it must be one of the other three that is formed by recombination. Analysis of a full sequence survey of the same region in a subset of these individuals (S. Alonso and J.A.L. Armour, manuscript in preparation), incorporating additional information from other sites, suggests that haplotype B is the result of a –8CT substitution on an A background and therefore that either C or D is formed by recombination or recurrent mutation. Although the impression that haplotype C has undergone major expansion may be partly due to bias towards the (mainly Eurasian) sample, complementary work using full sequence sampling of this region (S. Alonso and J.A.L. Armour, manuscript in preparation) also supports population expansion associated with this lineage. For the network as a whole, further information on likely haplotype relationships can be gained by looking at the minisatellite variant repeat (MVR) maps associated with each substitutional haplotype; the full data set is available from http://www.nott.ac.uk/pdzjala/205haplo.html .
|
|
MVR associations
In producing the 5' flanking haplotypes simple length variation of the minisatellite array was used with allele-specific PCR to deduce the phase. Each minisatellite array has further variation in the distribution of minisatellite variant repeats: the so-called ‘MVR map’ (14). This information can be integrated with each flanking substitutional haplotype to produce highly informative compound haplotypes. Since the minisatellite array mutates at high frequency, in the absence of recombination we would expect minisatellite alleles associated with the same flanking substitutional haplotype to have descended from a common ancestor and hence to form a monophyletic group. Therefore, we would expect each flanking substitutional haplotype to be associated with a group of similar, related MVR map structures.
For example, compound haplotype E, defined by the AC substitution at –382, is associated with a group of MVR maps of related structure, including those shown in Figure 4. The alleles generally differ by changes at the 3' end, as predicted for this array, at which 3' polarity of minisatellite mutation is already known (9–11). There are other MVR maps of similar structure associated with haplotype C, representing a related subset of lineages in one of which the original substitution at –382 took place.
|
The compound haplotypes consisting of the flanking haplotypes and associated MVR maps have added information to the system and thus allowed us to modify the initial network derived only from substitutional haplotypes. Compound haplotype G differs from the commonest Eurasian haplotype (C) by a G
C substitution at position –1627 and is found to be associated with a subset of MVR map structures of two types (Fig. 5), the longer of which are a subset of the variation associated with compound haplotype C. A –8TC ‘reversion’ on a haplotype G background could give rise to haplotype H, associated with a further subset of MVR maps (Fig. 5) found only in Melanesia in our sample. Similarly, haplotype I also differs from haplotype C by –8TC and associated MVR maps clearly identify this lineage as an offshoot of C, rather than examples of haplotype D which is identical at the flanking sites (Fig. 5). Other resolutions of the network preferentially involving recombinations are plausible and indeed recombinations involving –1627, –1583 and –1173 can account for all the loops in the network. However, although MVR maps do not clearly resolve the reticulation between haplotypes A, B, C and D, quantitative analyses of similarity between aligned MVR maps (12) support the origin of haplotype H from G by substitutional mutation or gene conversion as shown here, rather than by crossover from haplotype I. Although comparisons of MVR map structures using published measures of similarity (12) give only marginal support to G (over I) as the progenitor of H, similarity measures allowing (singly) gapped alignments strongly favour haplotype G. Haplotype D is associated with an MVR map similar to some MVR maps associated with haplotype J (Fig. 3) suggesting that D may represent an intermediate between haplotypes A and J.
|
Overall, integration of MVR maps into the network shows suggests there have been at least three and possibly four changes in humans at position -8: between haplotypes A and B, C and I, G and H and (possibly) C and D. For simplicity, Figures 2 and 3 show the origins of haplotypes H and I as –8T
C substitutions; the A–B–C–D reticulation is left unresolved in Figure 2, but haplotype C is shown as a recombinant between B and D in Figure 3. Is recurrent change at this position due to recombination or recurrent substitution? Simple crossover is not a good explanation of (for example) the origin of haplotype H, as the –8TC change is placed between characteristic derived states of –1627 and a distinctive subset of minisatellite arrays. Furthermore, no corresponding reciprocal haplotypes are observed. Gene conversion is a possible explanation, but if so does not appear to be operating at high frequency at other positions studied, suggesting that any gene conversion may preferentially involve the region around –8. However, previous studies have specifically implicated the other extremity (the 3' end) of the MS205 array as a conversion hotspot (9–11), though it is possible that the 3' end of the flanking DNA (i.e. the 5' end of the repeat array) is also preferentially subject to conversion. It is also possible that these changes at –8 are due to recurrent and revertant base substitutions. If so, this observation of four mutations at a single position is a significant excess (P = 0.005) and suggests that –8 (part of a CG doublet) may be a hypermutable base. It is noteworthy that the same substitution CT at –8 has occurred independently in humans and chimpanzees, though in the absence of extensive data on substitutional polymorphism for chimpanzee nuclear DNA it is difficult to estimate how often one would expect the same site to be polymorphic in both humans and chimpanzees by chance. Hypermutability of the CG doublet giving rise to multiple changes at –8 may thus be an explanation for the origins of haplotypes such as C, D, H and I without invoking gene conversion.
Population histories
Studying the MVR maps associated with each substitutional haplotype allows in principle the prediction of the age of a lineage: if minisatellite mutation rates are equal, older lineages will have more associated diversity. Slatkin and Rannala (15) used haplotype frequency and associated variation together with mutation rate to determine the maximum likelihood estimate for the age of a lineage, incorporating other parameters such as the population growth rate and selection coefficient. We have used this method to estimate dates for some lineages found in our sample in Europeans, or in Eurasians, but not in Africans.
Although mutation rates differ between different MS205 minisatellite alleles, small pool PCR (SP-PCR) has been used to directly measure the male germline mutation of different MS205 alleles (11; J.A.L. Armour, unpublished data) and appropriate mutation rates deduced from these experiments were used in the simulations.
For example, haplotype E (Fig. 4) has been observed throughout Eurasia in our sample but not in Africa. Within the 552 Eurasian chromosomes sampled we have observed 52 examples of haplotype E. Of these, 50 MVR-mapped examples of haplotype E are associated with 33 different MVR maps (Fig. 4). From SP-PCR work the male germline mutation rate for MVR maps associated with haplotype E is 0.6%. A simulation using these data and assuming neutrality and a population growth rate of 0.005 (16,17) gives a maximum likelihood estimate for the age of lineage E as 2420–2630 (95% CI) generations. Assuming a generation time of 20 years this is a lineage age of 48 400–52 600 years.
Simulations were carried out to obtain maximum likelihood estimates for lineages E, F, G and H (Table 3). Estimates of the age of lineages can be heavily dependent on the assumed demographic model and estimates of its component parameters (18). For this reason we have also shown estimated ages using growth rates ranging between 0.0025 and 0.008 (Table 3). The parameters in the simulations shown in Table 3 can be found at (http://www.nott.ac.uk/pdzjala/205haplo.html ).
|
Comparison with results of coalescent analysis of DNA sequence
The oldest Eurasian specific lineage (haplotype G) is 52 000–66 000 years old (Table 3), consistent with a recent expansion of Eurasian populations from an African source (19). Coalescent analysis of a full sequence survey of the same region gives a similar estimate for the age of this lineage, 68 800–77 200 years (S. Alonso and J.A.L. Armour, manuscript in preparation). However, the age estimates from the two studies for haplotype E and F differ greatly, with the estimates from the coalescence study of sequence diversity being consistently younger. Haplotype G, for which there is greatest concordance in age between the two studies, was observed seven times in the sequence survey, whereas haplotypes E and F were both observed only once in the sequence survey.
The comparison between our survey and the coalescent study (S. Alonso and J.A.L. Armour, manuscript in preparation) suggests that more extensive sampling, to give more precise estimates for the frequencies of particular haplotypes in our survey, combined with inclusion of information on allelic diversity within a lineage, may make the lineage ages estimated here more reliable. Thus, although lineage diversity methods require short regions with high information content and incorporate assumptions about demographic parameters, they may represent a rich source of data for the inference of population histories complementary to methods based only on coalescence analysis of DNA sequence.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTS AND DISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Genomic DNA samples
DNA samples from 389 individuals of the following origins were typed: 49 North European (CEPH panel), 30 Finnish, 10 Japanese, 9 Melanesian, 10 Surui, 10 Kenyan, 10 West African, 18 Zimbabwean (12), 25 British, 129 Basque, 24 Castillian (13); 14 Valencian (a kind gift from J.A. Montoro); 41 German (from Münster; a kind gift from M. Brinkworth); 5 Biaka Pygmies and 5 Mbuti Pygmies (obtained from the Coriell Cell Repository). Primate DNA samples were obtained from European Collection of Cell Cultures (ECACC).
Sequencing the MS205 5' flanking region from non-human primates
The MS205 5' flanking region was amplified in two overlapping amplicons, using primers 309.3 and 205Q and 205P and 205JB (9) (309.3, 5'-tctcctgactgacacggctcgg-3'; 205JB, 5'-ccggaggcggggtgtgtg-3'), and sequenced using the Big Dye cycle sequencing method (ABI Prism; PE Biosystems).
Typing the 5' flanking polymorphisms
PCR primers 309.3 and 205JB were used to amplify the MS205 5' flanking region from 10 ng of genomic DNA over 32 cycles (95°C for 1 min, 70°C for 1 min, 72°C for 3 min), in the buffer described by Jeffreys et al. (20). These PCR products were denatured with alkali and used to produce duplicate dot blot filters, which were probed with ASOs for each allelic form at each polymorphic position. Allele-specific PCR primers which differed at the 3' base were used to specifically amplify across the repeat array from each heterozygous position in genomic DNA in order to deduce the phase (9).
The substitutional polymorphisms (except –565) create or destroy sites for commercially available restriction enzymes (9,11). An Eco01091–Fnu4HI double digest can be used as an assay for –1173. The G form at –1583 creates a site for NheI. The restriction enzyme Cac8I can be used to assay –1627 (G/C). Therefore genotypes could be confirmed, where necessary, by restriction enzyme cleavage of a PCR product.
Further details of PCR primers or restriction enzyme assays are available on request. For any substitutional haplotypes observed only once, results were thoroughly checked and in some cases repeated or restriction digest assays carried out to confirm genotypes. If identical or very similar MVR maps were observed to be associated with different substitutional haplotypes then the results were again checked in the same way.
Dating lineages
The program bdmc21 (http://allele.bio.sunysb.edu ) was used to date individual lineages on the network. MS205 is assumed to be located in a selectively neutral region, therefore the growth rate and selection coefficient can be assumed to be entirely dependent on the exponential growth rate of the population. Based on published estimates (16,17), we have used exponential growth rates of 0.0025–0.008 for the Eurasian populations represented in our sample.
|
ACKNOWLEDGEMENTS |
|---|
We thank Martin Brinkworth, Keiji Tamaki, Yoshi Katsumata, Jose Americo Montoro, Mark Jobling, Alec Jeffreys and Mark Shriver for help with samples and Louise Williams for MVR mapping data. This study was supported by project grants to JALA from the Wellcome Trust (grants nos 047113/Z/96/Z and 054551) and by the award of a BBSRC studentship to E.J.R.
|
FOOTNOTES |
|---|
+ To whom correspondence should be addressed. Tel: +44 1159 249 924 ext. 42080; Fax: +44 1159 709 906; Email: john.armour@nott.ac.uk
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTS AND DISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
1 Hammer, M.F., Karafet, T., Rasanayagam, A., Wood, E.T., Altheide, T.K., Jenkins, T., Griffiths, R.C., Templeton, A.R. and Zegura, S.L. (1998) Out of Africa and back again: nested cladistic analysis of human Y chromosome variation. Mol. Biol. Evol., 15, 427–441.[Abstract]
2 Cann, R.L., Stoneking, M. and Wilson, A.C. (1987) Mitochondrial DNA and human evolution. Nature, 325, 31–36.
3 Tishkoff, S.A., Dietzsch, E., Speed, W., Pakstis, A.J., Kidd, J.R., Cheung, K., Bonne-Tamir, B., Santachiara-Benerecetti, A.S., Moral, P., Krings, M. et al. (1996) Global patterns of linkage disequilibrium at the CD4 locus and modern human origins. Science, 271, 1380–1387.[Abstract]
4 Castiglione, C.M., Deinard, A.S., Speed, W.C., Sirugo, G., Rosenbaum, H.C., Zhang, Y., Grandy, D.K., Bonne-Tamir, B., Pakstis, A.J., Kidd, J.R. and Kidd, K.K. (1995) Evolution of haplotypes at the DRD2 locus. Am. J. Hum. Genet., 57, 1445–1456.[Web of Science][Medline]
5 Harding, R.M., Fullerton, S.M., Griffiths, R.C. and Clegg, J.B. (1997) A gene tree for ß-globin sequences from Melanesia. J. Mol. Evol., 44, S133–S138.
6 Kaessmann, H., Heissig, F., von Haesler, A. and Pääbo, S. (1999) DNA sequence variation in a non-coding region of low recombination on the human X chromosome. Nature Genet., 22, 78–81.[Web of Science][Medline]
7 Jin, L., Underhill, P.A., Doctor, V., Davis, R.W., Shen, P., Cavalli-Sforza, L.L. and Oefner, P.J. (1999) Distribution of haplotypes from a chromosome 21 region distinguishes multiple prehistoric human migrations. Proc. Natl Acad. Sci. USA, 96, 3796–3800.
8 Royle, N.J., Armour, J.A.L., Webb, M., Thomas, A. and Jeffreys, A.J. (1992) A hypervariable locus D16S309 located at the distal end of 16p. Nucleic Acids Res., 20, 1164.
9 Armour, J.A.L., Harris, P.C. and Jeffreys, A.J. (1993) Allelic diversity at minisatellite MS205 (D16S309): evidence for polarized variability. Hum. Mol. Genet., 2, 1137–1145.
10 Jeffreys, A.J., Tamaki, K., MacLeod, A., Monckton, D.G., Neil, D.L. and Armour, J.A.L. (1994) Complex gene conversion events in germline mutation at human minisatellites. Nature Genet., 6, 136–145.[Web of Science][Medline]
11 May, C.A., Jeffreys, A.J. and Armour, J.A.L. (1996) Mutation rate heterogeneity and the generation of allele diversity at the human minisatellite MS205 (D16S309). Hum. Mol. Genet., 5, 1823–1833.
12 Armour, J.A.L., Anttinen, T., May, C., Vega, E.E., Sajantila, A., Kidd, J.R., Kidd, K.K., Bertranpetit, J., Pääbo, S. and Jeffreys, A.J. (1996) Minisatellite diversity supports a recent African origin for modern humans. Nature Genet., 13, 154–160.[Web of Science][Medline]
13 Alonso, S. and Armour, J.A.L. (1998) MS205 minisatellite diversity in Basques: evidence for a pre-Neolithic component. Genome Res., 8, 1289–1298.
14 Jeffreys, A.J., MacLeod, A., Tamaki, K., Neil, D.L. and Monckton, D.G. (1991) Minisatellite repeat coding as a digital approach to DNA typing. Nature, 354, 204–209.[Medline]
15 Slatkin, M. and Rannala, B. (1997) Estimating the age of alleles by use of intrallelic variability. Am. J. Hum. Genet., 60, 447–458.[Web of Science][Medline]
16 Pritchard, J.K., Seielstad, M.T., Perez-Lezaun, A. and Feldman, M.W. (1999). Population growth of human Y chromosomes: a study of Y chromosome microsatellites. Mol. Biol. Evol., 16, 1791–1798.[Abstract]
17 Thomson, R., Pritchard, J.K., Shen, P., Oefner, P.J. and Feldman, M.W. (2000). Recent common ancestry of human Y chromosomes: evidence from DNA sequence data. Proc. Natl Acad. Sci. USA, 97, 7360–7365.
18 Brookfield, J.F.Y. (1997) Importance of ancestral DNA ages. Nature, 388, 134.[Medline]
19 Pääbo, S. (1999) Human evolution. Trends Genet., 9, M13–M16.
20 Jeffreys, A.J., Wilson, V., Neumann, R. and Keyte, J. (1988) Amplification of human minisatellites by the polymerase chain reaction: towards DNA fingerprinting of single cells. Nucleic Acids Res., 16, 10953–10971.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  D. Garrigan, S. B. Kingan, M. M. Pilkington, J. A. Wilder, M. P. Cox, H. Soodyall, B. Strassmann, G. Destro-Bisol, P. de Knijff, A. Novelletto, et al. Inferring Human Population Sizes, Divergence Times and Rates of Gene Flow From Mitochondrial, X and Y Chromosome Resequencing Data Genetics, December 1, 2007; 177(4): 2195 - 2207. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. K. Shimada, K. Panchapakesan, S. A. Tishkoff, A. Q. Nato Jr, and J. Hey Divergent Haplotypes and Human History as Revealed in a Worldwide Survey of X-Linked DNA Sequence Variation Mol. Biol. Evol., March 1, 2007; 24(3): 687 - 698. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. A. Langdon and J. A.L. Armour Evolution and population genetics of the H-ras minisatellite and cancer predisposition Hum. Mol. Genet., April 15, 2003; 12(8): 891 - 900. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. L. Mountain, A. Knight, M. Jobin, C. Gignoux, A. Miller, A. A. Lin, and P. A. Underhill SNPSTRs: Empirically Derived, Rapidly Typed, Autosomal Haplotypes for Inference of Population History and Mutational Processes Genome Res., November 1, 2002; 12(11): 1766 - 1772. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Alonso and J. A. L. Armour A highly variable segment of human subterminal 16p reveals a history of population growth for modern humans outside Africa PNAS, December 14, 2000; (2000) 11244998. [Abstract] [Full Text]
|
|
|
|
|
|
S. Alonso and J. A. L. Armour A highly variable segment of human subterminal 16p reveals a history of population growth for modern humans outside Africa PNAS, January 30, 2001; 98(3): 864 - 869. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||