BRIEF COMMUNICATION
Population-specific patterns of linkage disequilibrium in the human 5q31region
G Luoni 1,{,J Forton 1,2,M Jallow 3,4,E Sadighi Akha 1,4,F Sisay-Joof 3,M Pinder 3,N Hanchard 1,2,M Herbert 2,M Kimber 1,R Mott 1,J Hull 2,K Rockett 1and D Kwiatkowski 1,2
1
Wellcome Trust Centre for Human Genetics,Oxford,UK;2University Department of Paediatrics,Oxford,UK;3MRC Laboratories,Fajara,The Gambia
Linkage disequilibrium across the human genome is generally lower in West Africans than Europeans.However in the 5q31region,which is rich in immune genes,we find significantly more examples of apparent nonrecombination between distant marker pairs in West Africans.Much of this effect is due to SNPs that are absent in Europeans,possibly reflecting recent positive selection in the West African population.
Genes and Immunity advance online publication,28July 2005;doi:10.1038/sj.gene.6364250
Introduction
How patterns of linkage disequilibrium (LD)vary between populations of different ethnic origin is highly topical,for two main reasons.First,populations exposed to different environmental pressures might be compared to detect potential signatures of recent selection.1Second,variation in LD structure is integral to the problem of defining haplotype-tagging strategies that are transfer-able across different populations in disease association studies.2
A dense concentration of immune genes is located in the human 5q31region,whose pattern of LD in individuals of European ancestry is well documented and has been described as a set of contiguous haplotype blocks.3
As the 5q31region is a susceptibility locus for parasitic infections including malaria and schistosomiasis,4,5we compared the haplotypic structure across the region in two populations that differ greatly in their exposure to these diseases.
The sample consisted of 128unrelated West African chromosomes and 128unrelated European chromo-somes.
The Th2cytokine cluster in the human 5q31region is known to be under coordinate regulation and several important enhancers and repressors have been defined at locations geographically distinct from the genes them-selves.6,7We therefore examined long-range haplotype structure rather than limiting comparison to the variation at each individual gene locus.We studied a 656kb
segment of the 5q31region containing 13genes including IL3,CSF2,IRF1,IL5,IL13and IL4(Figure 1).
We found that although there were generally higher levels of LD in the European population,more examples of long-range LD were seen in the West African population.We explore the possibility that long-range LD with high allele frequency may indicate recent positive selection and interrogate the SNPs in the region using the concept of relative extended haplotype homo-zygosity (rEHH).
Results and discussion
The sample comprised 128unrelated West African and 128European chromosomes,based on genotyping 32family trios in each population.The region was a 656kb segment containing 13genes including IL3,CSF2,IRF1,IL5,IL13and IL4.Of the 98SNPs analysed,89were found to be polymorphic in both populations and nine in West Africans only (see Figure 1for a map of the region;Supplementary Table A for a list of all SNPs studied and Supplementary Figure A for the distribution of minor allele frequencies for all SNPs).
The patterns of LD in each population are summarized in Figure 2and can be queried at https://www.doczj.com/doc/7717252575.html,/pub/002.Analysis of r 2,which is a simple measure of association between two polymorphisms,showed gen-erally higher levels of LD in Europeans than West Africans.However,we noted that high values of D 0(which indicate little evidence of recombination)were mostly concentrated in islands of neighbouring markers in Europeans,whereas in West Africans high D 0values had a more scattered distribution.
Paradoxically,we found that occurrences of an absolute value of D 0?1at a range of 4200kb were significantly more common in West Africans than Europeans (Table 1;odds ratio 2.6,P o 10à7by chi-
Received 9May 2005;revised 8June 2005;accepted 8June 2005
Correspondence:D Kwiatkowski,University Department of Paediatrics,Wellcome Trust Centre for Human Genetics,Roosevelt Drive,Oxford,Oxon,OX37BN,UK.E-mail:dominic@https://www.doczj.com/doc/7717252575.html, {
Deceased 4
These authors made an equal
contribution.
Genes and Immunity (2005),1–5
&2005Nature Publishing Group All rights reserved 1466-4879/05$30.00
https://www.doczj.com/doc/7717252575.html,/gene
squared test).This does not appear to be an artefact of allele frequency 8because the difference remains signifi-cant if we exclude SNPs with minor allele frequency (m.a.f.)p 0.1(odds ratio 3.1,P ?5?10à4).If we focus on pairs of SNPs that show no evidence of recombination (D 0?1)where both SNPs have m.a.f.40.1,they are much more likely to span a distance of 4200kb in West Africans than in Europeans (odds ratio 7.1,P o 10à7).As a different way of excluding possible artefacts due to low allele frequency and sample size,we also considered the credibility interval of absolute D 0,determined by computing the likelihood of the given pairwise marker observations as a function of D 0,and inferring the range of credible absolute D 0values.As described in Table 2,the likelihood density function for
each pairwise comparison was plotted for D 0from à1to 1,and we then identified the range of values that encompassed (for example)95%of the area under the curve.As shown in Table 2,instances of high LD between SNP pairs were significantly more likely to span a distance of 4200kb in West Africans than in Europeans using either of the following criteria for absolute D 0values:nominal value ?1and 95%credibility interval 40(West Africans vs Europeans,odds ratio 5.4,P o 10à9);or alternatively,nominal value ?1and 70%credibility interval 40.4(odds ratio 4.3,P o 10à9).
Considerably fewer instances of long-range high D 0values are observed if we exclude SNPs that are monomorphic in Europeans:for occurrences of D 0?1at a range of 4200kb for SNPs of m.a.f.40.1,the
odds
Figure 2LD between SNPs across the 5q31region,where the vertical axis represents all SNPs typed,and black dots represent pairs of SNPs with high LD.(a )West African sample showing instances of r 240.3;(b )West African sample showing instances of absolute D 040.9;(c)European sample showing instances of r 240.3;(d)European sample showing instances of absolute D 040.9.A detailed interactive version of this figure may be viewed at o https://www.doczj.com/doc/7717252575.html,/pub/0024
.
Patterns of LD in the human 5q31region
G Luoni et al
2
Genes and Immunity
ratio comparing West Africans to Europeans falls to 1.7(not significant;Table 1).Thus,much of the long-range LD that we observe in the African sample appears to arise from African-specific polymorphisms.
The combination of long-range LD with high allele frequency is indicative of recent positive selection,because if an allele has preserved its ancestral haplotypic relationships,this implies that it is relatively young.9,10For each SNP ,we attempted to assess whether the minor allele had preserved ancestral relationships with neigh-bouring SNPs,using the concept of relative extended haplotype homozygosity (rEHH 10).Figure 3shows that rEHH tends to decline with allele frequency but that one of the African-specific SNPs,an intronic SNP of IL4(rs2243261;m.a.f.?0.16),exhibits much higher levels of rEHH than other SNPs of a similar frequency within this data set (other views can be generated at https://www.doczj.com/doc/7717252575.html,/pub/002).At present,we lack robust statistical methods to validate signatures of positive selection,but such data are highly suggestive.It is noteworthy that immune genes are densely concentrated in the 5q31region,which is a known susceptibility locus for malaria and schistosomiasis,4,5and that polymorphism in IL4has previously been associated with antimalarial antibody responses.11Clearly,it is possible that the huge burden of infectious disease that afflicts African populations has been a significant determinant of the population-specific pattern of LD observed in the 5q31region,and it will be
Table 1Instances of SNP pairs with a nominal value of absolute
D 0?1,comparing those found at a range of above or below 200kb in Europeans or West Africans
Population
D 0o 1
D 0?1o 200kb D 0?1X 200kb All SNPs Afr (98SNPs)3817540399Eur (89SNPs)3367414135Excluding minor Afr (79SNPs)291912537allele frequency p 0.1
Eur (76SNPs)257526411Excluding minor Afr (71SNPs)236510416allele frequency p 0.1and those monomorphic in Europeans
Eur (76SNPs)
2575
264
11
Table 2Instances of SNP pairs with a nominal value of absolute
D 0?1,together with 95%credibility interval 40or 70%credibility interval 40.4,comparing those found at a range of above or below 200kb in Europeans or West Africans Absolute D 0
Population o 200kb X 200kb Nominal value ?1Afr (98SNPs)540399Eur (89SNPs)414135Nominal value ?1Afr (98SNPs)316123and 95%credibility interval 40
Eur (89SNPs)38928Nominal value ?1and Afr (98SNPs)33414870%credibility interval 40.4
Eur (89SNPs)
396
41
As D 0values tend to be overestimated for rare alleles in conditions of low sample size,we supplemented our point estimates of D 0with a credibility interval,along the same general lines as followed by Gabriel et al .19Specifically,after estimating the haplotypes,we made pairwise comparisons between each possible combination of markers in a two by two table with one degree of freedom.We compute the probability of a single chromosome falling into the n 11category,for a specified hypothetical D 0,as
p 11?ena nB D 0tnA nB T=N 2for D 04?0
or
p 11?ena nb D 0tnA nB T=N 2for D 0o 0
where na ?n 21+n 22,nA ?n 11+n 12,similarly for nb and nB ,and N ?n 11+n 12+n 21+n 22?total number of chromosomes.The like-lihood of observing the given two-by-two table is computed as proportional to p 11n11áp 12n12áp 21n21áp 22n22,where p 11is the probability shown,and the other probabilities follow from the constraint that the individual allele frequencies are treated as fixed.The likelihood density function is plotted for D 0from à1to 1and extreme values of D 0are identified between which (for example)95%of the area under the curve is included.After accounting for the sign,the resulting limits on the absolute value of D 0are quoted as 95%credibility limits for that particular marker–marker compar-
ison.
Figure 3Relative extended haplotype homozygosity (rEHH)for all SNPs in the West African population.A high value for rEHH signifies a relative lack of diversity in the group of haplotypes carrying the minor allele when compared to the group carrying the major allele.The combination of high rEHH with high allele frequency is suggestive of positive selection.In a set of n SNPs,each individual SNP x has n-1values for rEHH,representing the different haplotypic segments extending out from SNP x to each of the n-1surrounding SNPs.All of these n-1values are plotted for each SNP x in Figure 3.Rs2243261is marked with an https://www.doczj.com/doc/7717252575.html,/pub/002provides tools to view the decay of rEHH with distance for each SNP .A full explanation of the method for calculating rEHH is described by Sabeti et al
10
Patterns of LD in the human 5q31region G Luoni et al
3
Genes and Immunity
interesting to see whether this is repeated in other regions of the genome.
These data have practical implications for the selection of informative markers for genetic association studies.Much of the long-range LD in the West African sample came from eight SNPs that had minor allele frequencies 40.1in this population,but were nonpolymorphic in the European sample.As Europeans generally have higher values of linkage disquilibrium between neighbouring markers,they show a stronger pattern of haplotype blocks.This is illustrated in Figure 4,where all the SNPs that we tested were parsed by a greedy algorithm using
Haploblock finder,with the criterion of D 040.8.12For Europeans,this yielded 20blocks and 95%of poly-morphic markers were situated in blocks of two or more.The corresponding figures in Africans were 48and 73%.To capture this block structure,we would need to type 43SNPs in Europeans and 70in Africans.Thus a haplotype block analysis suggests that the set of informative markers needed to capture the diversity of the region is much less in Europeans than in Africans,but this loses the information that we could get by considering the relationships that exist between distant markers.For example,if we use the ENTROPY algorithm,we find that 21SNPs in Europeans and 18SNPs in Africans can capture the full information content of the marker set.In other words,a haplotype block approach is a relatively economical way of defining tagging SNPs for a popula-tion where high LD is mostly observed at a relatively close range,but is a less economical solution in situations where high LD tends to be more dispersed,and the present data indicate that this may be particularly true of African populations.
Acknowledgements
This work was funded by a Marie Curie fellowship (GL),a Wellcome Trust research training fellowship (JF)and the Medical Research Council.
References
1Hinds DA,Stuve LL,Nilsen GB et al.Whole-genome patterns of common DNA variation in three human populations.Science 2005;307:1072–1079.
2Ahmadi KR,Weale ME,Xue ZY et al.A single-nucleotide polymorphism tagging set for human drug metabolism and transport.Nat Genet 2005;37:84–89.
3Daly MJ,Rioux JD,Schaffner SF et al.High-resolution haplotype structure in the human genome.Nat Genet 2001;29:229–232.
4Rihet P ,Traore Y,Abel L et al.Malaria in humans:Plasmodium falciparum blood infection levels are linked to chromosome 5q31–q33.Am J Hum Genet 1998;63:498–505.
5Marquet S,Abel L,Hillaire D et al.Genetic localization of a locus controlling the intensity of infection by Schistosoma mansoni on chromosome 5q31–q33.Nat Genet 1996;14:181–184.
6Ansel KM,Lee DU,Rao A.An epigenetic view of helper T cell differentiation.Nat Immunol 2003;4:616–623.
7Lee GR,Fields PE,Griffin TJ et al.Regulation of the Th2cytokine locus by a locus control region.Immunity 2003;19:145–153.
8Ardlie KG,Kruglyak L,Seielstad M.Patterns of linkage disequilibrium in the human genome.Nat Rev Genet 2002;3:299–309.
9Bamshad M,Wooding SP .Signatures of natural selection in the human genome.Nat Rev Genet 2003;4:99–111.
10Sabeti PC,Reich DE,Higgins JM et al.Detecting recent positive selection in the human genome from haplotype structure.Nature 2002;419:832–837.
11Luoni G,Verra F,Arca B et al.Antimalarial antibody levels and IL4polymorphism in the Fulani of West Africa.Genes Immun 2001;2:411–414.
12Zhang K,Jin L.HaploBlockFinder:haplotype block analyses.Bioinformatics 2003;19:
1300–1301.
Figure 4Haplotype tagging SNPs in the 5q31region.Haplotypes of the 128parental chromosomes from each population were determined from the pedigree data,using the PHASE algorithm to resolve ambiguities that arose in 8%of European and 9.5%of African phase assignments.16To illustrate the haplotype block approach for selecting tagging SNPs,we used the greedy algorithm provided by HaploBlockFinder 12(closed circles).To illustrate an unstructured method of selecting tagging SNPs,we used ENTROPY which determines the information content of each SNP without consideration of block structure 17,18(open squares).The haplotype-block-based method identified 43tagging SNPs in Europeans and 70in West Africans,while the unstructured method identified 21tagging SNPs in Europeans and 18in West
Africans.
Patterns of LD in the human 5q31region
G Luoni et al
4
Genes and Immunity
13Rioux JD,Daly MJ,Silverberg MS et al.Genetic variation in the 5q31cytokine gene cluster confers susceptibility to Crohn disease.Nat Genet2001;29:223–228.
14Noguchi E,Nukaga-Nishio Y,Jian Z et al.Haplotypes of the50 region of the IL-4gene and SNPs in the intergene sequence between the IL-4and IL-13genes are associated with atopic asthma.Hum Immunol2001;62:1251–1257.
15Howard TD,Whittaker PA,Zaiman AL et al.Identification and association of polymorphisms in the interleukin-13gene with asthma and atopy in a Dutch population.Am J Resp Cell Mol Biol2001;25:377–384.16Stephens M,Smith NJ,Donnelly P.A new statistical method for haplotype reconstruction from population data.Am J Hum Genet2001;68:978–989.
17Ackerman H,Usen S,Mott R et al.Haplotypic analysis of the TNF locus by association efficiency and entropy.Genome Biol 2003;4:R24.
18Mott R.Marker selection by maximum entropy.http:// https://www.doczj.com/doc/7717252575.html,/B rmott/SNPS/2003.
19Gabriel SB,Schaffner SF,Nguyen H et al.The structure of haplotype blocks in the human genome.Science2002;296: 2225–2229.
Supplementary information accompanies the paper on Genes and Immunity website
(https://www.doczj.com/doc/7717252575.html,/gene).
Patterns of LD in the human5q31region
G Luoni et al
5
Genes and Immunity
4A广告公司职位中英文对照 4A公司（综合性代理广告公司）部门架构与岗位职称Management/Administration管理/行政部门 Chief Executive Officer 首席行政长官（常用简称：CEO） Managing Director 董事总经理（常用简称：MD） General Manager 总经理（常用简称：GM） HI & Admin Manager 人事/行政经理 HI Supervisor 人事督导 Admin Assistant 行政助理 Information Technology 信息技术部门 Information Technology Manager 信息技术经理 Information Technology Administrator信息技术管理员 Information Technology Support信息技术支持 Account Management客户服务部门 Director of Client Services 客户服务部总监 Group Account Director客户事务群总监（常用简称：GAD）Senior Account Director高级客户总监（常用简称：SAD） Account Director客户总监（常用简称：AD） Associate Account Director助理客户总监（常用简称：AAD）Account Manager客户经理（常用简称：AM） Account Manager-Freelance 兼职客户经理 Senior Account Executive 高级客户执行（常用简称：SAE）
以下是我整理的国内主要4A广告公司和媒体购买公司及其服务客户名单，请各位广告圈的朋友多多补充更新，共同进步 星传媒体Starcom、李奥贝纳广告Leo Burnett： 箭牌系列产品，丹麦曲奇，惠氏产品，西联汇款，通用汽车，菲亚特汽车，美标洁具，湾仔码头食品，李宁，玛氏食品，P&G，美赞臣（安儿宝），南孚电池，三菱手机，中国电信－网络， 麦肯光明McCann-Erickson： 通用磨房，欧莱雅，纳贝斯克食品，摩托罗拉手机，雀巢，强生，高露洁，黑人牙膏，广发银行，多美滋食品，吉列，UPS，美宝莲，国泰航空，广州移动广旭广告、电通广告Dentsu： 佳能，小护士，松下，西铁城，厦新手机，东芝，东陶，丰田，花王，伊莱克斯，联想，三得利饮料，三菱电机，优衣裤休闲服 东方日海广告： 胡标，奥林帕斯，柯尼卡，索尼，深圳联通，科键，友邦保险－深圳分公司，中国银行深圳分行，朝日啤酒，康佳电视， 传立媒体Mindshare、智威汤逊J.Walter Thompson、奥美广告Ogilvy &Mather、灵立媒体、灵智大洋Euro Rscg： 金佰利，荷氏薄荷糖，联合利华，舒适剃须刀，荷兰子母奶，必胜客，多乐士，柯达，尤妮佳,百胜餐饮，三得利，ICI，雪铁龙，沃尔沃，戴尔，微软，卡夫，FordMotor,肯德基，耐克，老蔡酱油，西门子手机，IBM,上海大众，帝比亚斯钻石Debeers，立顿，百事可乐，雅居乐，雅芳，派斯油漆， 实力媒体Zenith Media、盛世长城Saatchi & Saatchi、达彼思广告BatesAdvertising： 阿尔卡特，光明，安利，惠泉啤酒，宜家，中意人寿，诺华制药，GE，瑞士军刀，金盛人寿，百威啤酒，可口可乐，SC强生，罗氏，汇丰银行，CK,索
MD〈Managing Director〉——广告公司内的最高统帅，中文惯例译为总经理。 CD〈Creative Director〉——创作总监。CD的前身，不是撰稿人便是美术设计，因为积累了丰富的经验，并有优异的创作成绩而成为督导。 ACD〈Associated Creative Director〉——副创作总监。 Account Director——客户服务总监〈客户部经理〉。 Media Director——媒介总监〈媒介部经理〉。 AD〈Art Director〉——美术指导〈在创作部可以独挡一面执行美术指导工作的美术监督〉。Copywriter——撰稿人。 AE〈Account Executive〉——客户服务人员。 Visualizer——插图家〈插画师〉。 Studio Manager——画房经理。 Finish artist——画师。 AE〈Account Executive〉——预算执行者，负责广告代理商和广告主之间的一切有关业务，观念，预算，广告表现之联系。 CF〈Commercial Film〉——广告影片。 OS〈Omt Sound〉——广告影片中的旁白。 POP〈Point of Perchase advertising〉——购买时点的广告，大部分消费者买东西时在商店因此也称店头广告。 SP〈Sales Promotion〉——促销活动。 CIS〈Corporated Identity System〉——企业识别系统。 CI〈Corporated Identity〉——企业识别。 CI 与CIS之区分，在美国原来只购CI，日本加上“S”也就是把CI加以组织化，包括EI，VI，BI，SI，成为系统。 EI〈Environment Identity〉——环境识别。 CS〈Customer Satisfaction〉——顾客满意度。 SI〈Store Identity〉——商店识别。 USP〈Unique Saling Point〉——独特性的销售主张。 DM〈Direct Mail〉——广告函件，直接信函，广告主将印刷品以邮寄方式直接寄给特定对象 DM〈Direct Marketing〉——直接行销，直接与消费者接触的行销方式，如广告信函，人员直接销售