Nature Genetics 38, 82 - 85 (2006)
Published online: 4 December 2005; | doi:10.1038/ng1695
Common deletions and SNPs are in linkage disequilibrium in the human genomeDavid A Hinds1, 2, Andrew P Kloek1, 2, Michael Jen1, Xiyin Chen1
& Kelly A Frazer11 Perlegen Sciences, Inc., 2021 Stierlin Court, Mountain View, California 94043, USA. 2 These authors contributed equally to this work.
Correspondence should be addressed to Kelly A Frazer kelly_frazer@perlegen.com Humans show great variation in phenotypic traits such as height, eye color and susceptibility to disease. Genomic DNA sequence differences among individuals are responsible for the inherited components of these complex traits. Reports suggest that intermediate and large-scale DNA copy number and structural variations are prevalent enough to be an important source of genetic variation between individuals1,
2,
3,
4,
5,
6,
7,
8. Because association studies to identify genomic loci associated with particular phenotypic traits have focused primarily on genotyping SNPs, it is important to determine whether common structural polymorphisms are in linkage disequilibrium with common SNPs, and thus can be assessed indirectly in SNP-based studies. Here we examine 100 deletion polymorphisms ranging from 70 bp to 7 kb. We show that common deletions and SNPs ascertained with similar criteria have essentially the same distribution of linkage disequilibrium with surrounding SNPs, indicating that these polymorphisms may share evolutionary history and that most deletion polymorphisms are effectively assayed by proxy in SNP-based association studies.
SNPs are the result of errors in DNA replication or repair that occurred once in human history and are shared among individuals by descent9,
10,
11. Very small common deletions and insertions, in the range of 1–5 bp, show strong linkage disequilibrium with common SNPs, which suggests that, although the mechanisms giving rise to them may differ, these polymorphisms share a similar evolutionary history12. It is well documented that diseases classified as genomic disorders, such as DiGeorge or velocardiofacial syndrome13,
14,
15,
16,  -thalassemia17,
18, Williams-Beuren syndrome19 and Charcot-Marie-Tooth disease type 1A20, result from recurring mutations involving large deletions, insertions and other genomic alterations21. These recurring mutations are the result of non-allelic homologous recombination events that occur between blocks of duplicated sequences (>95% sequence identity, >10 kb in length, and separated by 50 kb to 10 Mb)21. Here we and other concurrent reports22,
23 in this issue show that intermediate-length deletion polymorphisms contribute to common genetic variation in healthy individuals. Our report focuses on whether these common deletion polymorphisms are the result of single mutation events such as SNPs or are due to recurring mutational events such as those resulting in genomic disorders.
To identify human deletions, we used an array-based comparative genomics approach to examine 24 unrelated individuals (termed the 'Discovery Panel') obtained from the Polymorphism Discovery Resource24. Haploid hybridization data25 was used to identify genomic intervals showing a reduced hybridization signal in comparison to the reference human genome sequence, which was tiled on the array as described26 (Fig. 1). We detected 215 candidate deletions ranging from 70 bp to 10 kb by screening across roughly 600 Mb of genomic DNA. About half of the 600 Mb of sequence screened consisted of repetitive sequences that were not examined here. We estimate that one- to two-thirds of the 300 Mb of unique sequence was effectively assayed for deletions.
 | | |
For each putative deletion, primer pairs were designed to flank the predicted position of the deletion so that its presence could be detected by PCR as an amplification product smaller than that from non-deleted alleles. Putative deletions were examined by PCR using diploid genomic DNA from each of the 24 individuals in the Discovery Panel and were genotyped as homozygous for the reference sequence (+/+, one upper band), or heterozygous (+/-, two bands) or homozygous (-/-, one lower band) for the deletion (Fig. 1). We selected a subset of 100 deletions (Supplementary Tables 1 and 2 online) that were unambiguously confirmed on the basis of their PCR patterns for further analyses.
The 100 deletions map to 14 different chromosomes and range from about 70 bp to 7 kb with a median of 750 bp. The PCR-based genotype data (Supplementary Table 2) showed that 33 of the deletions were singletons in the 24 Discovery Panel individuals, whereas 41 were common in these individuals with an allele frequency of 10% or greater. We found that 43 deletions overlap transcripts and two span exons: deletion 20 results in loss of a 162-bp coding exon in the tumor suppressor gene MTUS1; and deletion 23 results in loss of an 89-bp exon from the 5' untranslated region of MS4A1 (previously known as CD20), a B cell surface antigen.
Several studies1,
2,
5,
6 report the identification of common copy number and structural variants in the human genome. Most of the polymorphisms reported in those studies are larger than 10 kb. The polymorphisms in one study1 include a few smaller variants but, on the basis of the coordinates reported, none of these overlaps with the deletions found in our study. Here, 6 of the 100 intermediate-length deletions (deletions 10, 15, 22, 35, 60 and 91) map within genomic sequences flanked by local duplications reported as 'hot spots' for rearrangement by non-allelic homologous recombination2,
27.
To evaluate the population frequency of deletion polymorphisms as compared with SNPs, we selected 100 SNPs for each deletion in our set with similar ascertainment characteristics (see Methods) and compared their allele frequencies in the haploid Discovery Panel data (Fig. 2). Of the 100 deletions, 67% showed minor allele frequencies (MAFs) of 10% or less. Of the 10,000 SNPs in the analysis, 52% showed MAFs in this range. Although the deletions thus seem to be slightly overrepresented for rare variants as compared with SNPs in these samples, there are many differences in how the two types of variant were identified and this result may reflect an ascertainment bias against the identification of rare SNP alleles in our data.
| | |
We used PCR to determine genotypes for each of the 100 deletions in a larger group of 71 ethnically diverse individuals that have been used to examine whole-genome patterns of common SNP variation in humans28 (Fig. 1 and Supplementary Tables 1 and 3 online). The 71 samples (termed the 'Diversity Panel') were obtained from the Human Variation Collection of the Coriell Cell Repositories and comprise individuals from three populations: 23 African Americans, 24 European Americans and 24 Han Chinese. We found that 84 of the 100 deletions were polymorphic in the 71 Diversity Panel individuals. As observed with SNP data28, the African American samples showed more diversity, as measured by relatively higher numbers of segregating and private deletions, in comparison to the European American and Han Chinese samples (Table 1).
| | |
Because the 71 individuals in the Diversity Panel have been used to construct a linkage disequilibrium map of 1.6 million common human SNPs28, an analysis of the relationship between SNPs and deletion polymorphisms present in these individuals was possible. Common deletions segregating with an allele frequency of  10% were identified from each of the three populations in the Diversity Panel (Table 1). For each common deletion, we determined the maximum square of the correlation coefficient (r2) with any genotyped common SNP located within 50 kb on either side of the deletion. We then determined the same statistic for a set of SNPs with ascertainment criteria similar to those of the deletions. In each of the three populations, deletion polymorphisms showed comparable linkage disequilibrium to SNPs (Fig. 3). There was no apparent association between deletion length and linkage disequilibrium, but the power to detect a relationship in our sample size was relatively low. These results suggest that intermediate-length deletions and SNPs have a similar evolutionary history and that most deletion polymorphisms are effectively assayed by neighboring SNPs.
| | |
We aligned our deletion loci to chimpanzee genomic sequences to determine which allele (deleted or non-deleted) reflects the ancestral sequence. Out of 95 deletion polymorphisms that could be aligned unambiguously to chimpanzee sequences, we observed only one full deletion and one partial deletion in the chimp, indicating that the human deletion was nearly always the derived allele. Given that our approach cannot detect insertions relative to the human reference genome, we cannot actually determine from these data whether deletions or insertions are more frequent events.
We inspected the nucleotide sequences contained in the deleted intervals to establish whether a molecular mechanism could be proposed to explain the origins of these rearrangements. We found that both unique and various repetitive sequences are present in the deleted sequences. For most repeat classes, the sequence composition of the deletions that we identified was similar to genome-wide averages. However, the deletions had a lower content of long interspersed elements than the genome average (8% versus  20%). Rigorous analysis of the repeat content of the deletions is difficult because the presence of repeats affects our ability to detect and determine accurately the endpoints of the deleted interval. Thus, underrepresentation of long interspersed elements may reflect a bias against the detection of deletions containing longer repeat classes. The repeat structure of a few deletions suggested a mechanism for loss of the interval; for example, similar repeats were sometimes found at each end of the interval. However, most deletions did not contain this pattern and thus our analysis neither implicated a particular class of sequences nor indicated an obvious mechanism that gives rise to these rearrangements.
Our platform enables us to evaluate most unique sequences in the human genome: we can amplify nearly 94% of all human sequences and examine even short stretches ( 60 bp) of relatively non-redundant sequence flanked by repetitive elements. Thus, we have effectively queried a representative fraction of the nonrepetitive sequence in the human genome and consider that the properties of the set of intermediate-length deletions identified in our study are representative of deletions in typical genomic contexts. Notably, unusual local sequence structures exist across the genome that create regions of genomic instability and result in recurrent rearrangements. For example, large (>10 kb) insertion/deletion polymorphisms associated with local duplications (sharing >95% identity and separated by 50 kb to 10 Mb) result from recurrent events1,
2,
5,
6,
27,
29. These large rearrangements are responsible for several known human disorders and may complicate the interpretation of SNP-based association studies. Although we saw little evidence of recurrent deletion events, we did not focus on genomic intervals located between local duplications. Thus, our coverage of such regions is necessarily limited and we cannot comment definitively on properties of common intermediate-length deletions there. In addition, it has been suggested that regions of the genome that are rich in transposon-like and other repetitive sequences are predisposed to recurrent genomic rearrangements and deletions21,
29. Although intervals of the genome with very high transposon content are likely to be too repetitive for us to assay effectively, such intervals comprise only a small fraction of the genome.
The set of common intermediate-length deletions identified here has linkage disequilibrium patterns similar to SNPs, indicating that these polymorphisms share a similar evolutionary history and suggesting that most intermediate-length deletions, like SNPs, arose once in human history. High linkage disequilibrium with nearby SNPs suggests that most of these deletions are effectively assayed by proxy in SNP-based association studies, consistent with previous results for short insertion/deletion polymorphisms12. On the basis of the fraction of the genome examined and the technical limits of our study, we estimate there are several thousand intermediate insertion/deletion polymorphisms in the human genome, suggesting that they represent an important component of common genetic variation and are likely to contribute to phenotypic variation in complex traits.
Methods
Human DNA samples.
The Discovery Panel consisted of 24 ethnically diverse individuals from the Polymorphism Discovery Resource24. The Diversity Panel consisted of 71 ethnically diverse individuals, including 24 European American samples, 23 African American samples and 24 Han Chinese samples, obtained from the Coriell Cell Repositories that had previously been used for large-scale SNP genotyping28.
High-density oligonucleotide array design and identification of deletions.
The high-density arrays, DNA labeling, hybridization and signal detection have been described26. Eight unique oligonucleotides, each 25 bases in length, were used to interrogate each base evaluated. Conformance was calculated in 30-bp windows by comparing Perlegen base calls with those expected from the human genome reference sequence. For example, if 25 out of 30 base calls matched, conformance was 83% for that window. We plotted and viewed conformance in 30-bp windows with 20-bp shifts. In Figure 1a, the conformance data points are connected and the area below the line is filled in red to facilitate observation of conformance trends. Regions of unexpectedly low conformance were selected as potential deletion loci, especially if other individuals showed relatively high conformance in the same segment. We specifically selected intervals that did not correspond to boundaries of the long-range PCR amplicons used to prepare the DNA samples for hybridization. Deletion positions and lengths were estimated from the conformance data, and sequences contained in the deletion intervals in the Discovery Panel data were used to identify their coordinates in the current human genome assembly (Build 35, May 2004) using the University of California Santa Cruz (UCSC) BLAT web server.
Comparison of deletion polymorphism and SNP allele frequencies.
The expected allele frequency distribution for a set of polymorphisms depends on the number of chromosomes successfully assayed during polymorphism discovery. Thus, to make a fair comparison of deletion and SNP allele frequencies, we used the same haploid hybridization data for SNP and deletion discovery25 and compared SNPs and deletions that were ascertained from the same number of samples. Missing call rates were much lower in the deletion data, however, because of the availability of more redundant hybridization data for each variant. To adjust for this difference, for each deletion we identified 100 SNPs at random that had been ascertained by using the same number of high-quality genotype calls on the haploid samples in the Discovery Panel for a total of 10,000 comparison SNPs. We then computed minor allele frequencies in the Discovery Panel on the basis of both the hybridization data for the deletions and the genotype calls for these comparison SNPs.
PCR verification of deletions.
We used PCR to verify deletions as follows: 11 ng of genomic DNA was amplified with primer pairs (each primer at 0.2  M), 0.29 U of EpiTaq (Epicentre), 0.1 g of TaqStart antibody (Becton Dickinson), 0.31 l of antibody buffer, 2.25 mM dNTPs, 0.14 l of Tricine (1 M), 0.17 l of dimethyl sulfoxide, 22 mM Tris-HCl (pH 9.1), 1.2 mM MgCl2, 6 mM ammonium sulfate, 2.6 mM KCl and 0.25 l of 10  MasterAmp PCR enhancer (Epicentre) in a volume of 6 l. Thermocycling was done with a 9700 cycler (Perkin-Elmer) as follows: initial denaturation at 94 °C for 3 min; 10 cycles of 94 °C for 2 s and 64 °C for 15 min; 28 cycles of 94 °C for 2 s and 64 °C for 15 min, with a 20-s increase per cycle; and a final extension at 62 °C for 60 min.
Primer design.
Primers for laboratory validation of suspected deletions were designed by using Oligo Primer Analysis Software Version 6.57. Primers were designed as 32-nucleotide oligomers by using the default 'high' search stringency and an oligomer Tm range from 72 to 88 °C. Oligo 6.57 was set to automatically choose highly specific primer pairs from the selected chromosomal regions, while eliminating pairs with the potential to self-prime or form hairpins and pairs containing common sequence repeats. Primers were selected to generate amplicons such that putative deletions would be easily identifiable by agarose gel electrophoresis.
Analysis of linkage disequilibrium.
To determine the extent to which deletion polymorphisms are effectively assayed by nearby SNPs, we computed r2 between common deletions and SNPs located within 50 kb that had previously been genotyped in the same Diversity Panel28. To compute r2, we used an expectation maximization algorithm to estimate pairwise haplotype frequencies from diploid genotypes30. We compared these results against linkage disequilibrium data for common SNPs with similar ascertainment. For each deletion, we selected 10 SNPs at random that had been ascertained with the same number of haploid samples from the Discovery Panel for a total of 1000 comparison SNPs. We then computed r2 between common SNPs from this set and other available genotyped SNPs located within 50 kb.
Chimpanzee genomic comparison.
For the intervals surrounding each deletion, we determined the best alignment to the November 2003 assembly of the chimpanzee genome, using the chimp chained alignment track of the UCSC Genome Browser. From this best alignment, we determined whether any portion of the human deletion was also deleted in the chimp.
Repeat content of deletions.
We determined the repeat content of each deletion interval by using the RepeatMasker track from the UCSC Genome Browser. For 13 deletions in which one end could not be determined from the hybridization data, we used our estimate of the deletion length from gel data to determine an approximate position. For 12 deletions in which both ends were uncertain, we chose arbitrary endpoints consistent with the hybridization and gel data. We then tabulated the number of bases in the deletion intervals contained in repeats by repeat class.
URLs.
The UCSC Genome Browser is available at http://genome.ucsc.edu. The UCSC BLAT web server is available at http://genome.ucsc.edu/cgi-bin/hgBlat. The website for the Human Variation Collection of the Coriell Cell Repositories is available at http://locus.umdnj.edu/ccr/.
Note: Supplementary information is available on the Nature Genetics website.
Received 30 June 2005; Accepted 6 October 2005; Published online: 4 December 2005.
REFERENCES
-
Tuzun, E.
et al. Fine-scale structural variation of the human genome. Nat. Genet. 37, 727–732 (2005). | Article | PubMed | ISI | ChemPort |
-
Sharp, A.J.
et al. Segmental duplications and copy-number variation in the human genome. Am. J. Hum. Genet. 77, 78–88 (2005). | Article | PubMed | ISI | ChemPort |
-
Stefansson, H.
et al. A common inversion under selection in Europeans. Nat. Genet. 37, 129–137 (2005). | Article | PubMed | ISI | ChemPort |
-
Singleton, A.B.
et al. -Synuclein locus triplication causes Parkinson's disease. Science 302, 841 (2003). | Article | PubMed | ISI | ChemPort |
-
Sebat, J.
et al. Large-scale copy number polymorphism in the human genome. Science 305, 525–528 (2004). | Article | PubMed | ISI | ChemPort |
-
Iafrate, A.J.
et al. Detection of large-scale variation in the human genome. Nat. Genet. 36, 949–951 (2004). | Article | PubMed | ISI | ChemPort |
-
Gonzalez, E.
et al. The influence of CCL3L1 gene-containing segmental duplications on HIV-1/AIDS susceptibility. Science 307, 1434–1440 (2005). | Article | PubMed | ISI | ChemPort |
-
Dhami, P.
et al. Exon array CGH: detection of copy-number changes at the resolution of individual exons in the human genome. Am. J. Hum. Genet. 76, 750–762 (2005). | Article | PubMed | ISI | ChemPort |
-
Cargill, M.
et al. Characterization of single-nucleotide polymorphisms in coding regions of human genes. Nat. Genet. 22, 231–238 (1999). | Article | PubMed | ISI | ChemPort |
-
Halushka, M.K.
et al. Patterns of single-nucleotide polymorphisms in candidate genes for blood-pressure homeostasis. Nat. Genet. 22, 239–247 (1999). | Article | PubMed | ISI | ChemPort |
-
Wang, D.G.
et al. Large-scale identification, mapping, and genotyping of single-nucleotide polymorphisms in the human genome. Science 280, 1077–1082 (1998). | Article | PubMed | ISI | ChemPort |
-
Bhangale, T.R.
,
Rieder, M.J.
,
Livingston, R.J.
&
Nickerson, D.A.
Comprehensive identification and characterization of diallelic insertion-deletion polymorphisms in 330 human candidate genes. Hum. Mol. Genet. 14, 59–69 (2005). | PubMed | ISI | ChemPort |
-
Burn, J.
,
Wilson, D.
&
Cross, I.
The clinical significance of 22q11.2 deletion. in Developmental Mechanisms of Heart Disease (eds. Clark, E.B., Markwald, R.R. & Takao, A.) 559–567 (Futura, New York, 1995).
-
Goodship, J.
,
Cross, I.
,
Liling, J.
&
Wren, C.
A population study of chromosome 22q11 deletions in infancy. Arch. Dis. Child. 79, 348–351 (1998). | PubMed | ISI | ChemPort |
-
Liling, J.
et al. Frequency and predictive value of 22q11 deletion. J. Med. Genet. 36, 794–795 (1999). | PubMed | ISI | ChemPort |
-
Tezenas Du Montcel, S.
,
Mendizabai, H.
,
Ayme, S.
,
Levy, A.
&
Philip, N.
Prevalence of 22q11 microdeletion. J. Med. Genet. 33, 719 (1996). | PubMed | ChemPort |
-
Higgs, D.R.
,
Old, J.M.
,
Pressley, L.
,
Clegg, J.B.
&
Weatherall, D.J.
A novel -globin gene arrangement in man. Nature 284, 632–635 (1980). | Article | PubMed | ISI | ChemPort |
-
Lauer, J.
,
Shen, C.K.
&
Maniatis, T.
The chromosomal arrangement of human -like globin genes: sequence homology and -globin gene deletions. Cell 20, 119–130 (1980). | Article | PubMed | ISI | ChemPort |
-
Greenberg, F.
Williams syndrome professional symposium. Am. J. Med. Genet. 6 (Suppl.), 85–88 (1990). | Article |
-
Nelis, E.
et al. Estimation of the mutation frequencies in Charcot-Marie-Tooth disease type 1 and hereditary neuropathy with liability to pressure palsies: a European collaborative study. Eur. J. Hum. Genet. 4, 25–33 (1996). | PubMed | ISI | ChemPort |
-
Inoue, K.
&
Lupski, J.R.
Molecular mechanisms for genomic disorders. Annu. Rev. Genomics Hum. Genet. 3, 199–242 (2002). | Article | PubMed | ISI | ChemPort |
-
Conrad, D.F.
,
Andrews, T.D.
,
Carter, N.P.
,
Hurles, M.E.
&
Pritchard, J.K.
A high-resolution survey of deletion polymorphism in the human genome. Nat. Genet., advance online publication 4 December 2005 (10.1038/ng1697). | Article |
-
McCarroll, S.A.
et al. Common deletion polymorphisms in the human genome. Nat. Genet., advance online publication 4 December 2005 (10.1038/ng1696). | Article |
-
Collins, F.S.
,
Brooks, L.D.
&
Chakravarti, A.A.
DNA polymorphism discovery resource for research on human genetic variation. Genome Res. 8, 1229–1231 (1998). | PubMed | ISI | ChemPort |
-
Patil, N.
et al. Blocks of limited haplotype diversity revealed by high-resolution scanning of human chromosome 21. Science 294, 1719–1723 (2001). | Article | PubMed | ISI | ChemPort |
-
Frazer, K.A.
et al. Genomic DNA insertions and deletions occur frequently between humans and nonhuman primates. Genome Res. 13, 341–346 (2003). | Article | PubMed | ISI | ChemPort |
-
Bailey, J.A.
et al. Recent segmental duplications in the human genome. Science 297, 1003–1007 (2002). | Article | PubMed | ISI | ChemPort |
-
Hinds, D.A.
et al. Whole-genome patterns of common DNA variation in three human populations. Science 307, 1072–1079 (2005). | Article | PubMed | ISI | ChemPort |
-
Shaw, C.J.
&
Lupski, J.R.
Implications of human genome architecture for rearrangement-based disorders: the genomic basis of disease. Hum. Mol. Genet. 13, R57–64 (2004). | Article | PubMed | ISI | ChemPort |
-
Weir, B.S.
Genetic Data Analysis II (Sinauer, Sunderland, Massachusetts, 1996).
Acknowledgments
We thank D. Cox for discussions and comments on the manuscript.
Competing interests statement:
The authors declare that they have no competing financial interests. |
