Copy Number Variation Sequencing for Comprehensive Diagnosis of Chromosome Disease Syndromes

Detection of chromosome copy number variation (CNV) plays an important role in the diagnosis of patients with unexplained clinical symptoms and for the identification of chromosome disease syndromes in the established fetus. In current clinical practice, karyotyping, in conjunction with array-based methods, is the gold standard for detection of CNV. To increase accessibility and reduce patient costs for diagnostic CNV tests, we speculated that next-generation sequencing methods could provide a similar degree of sensitivity and specificity as commercial arrays. CNV in patient samples was assessed on a medium-density single nucleotide polymorphism array and by low-coverage massively parallel CNV sequencing (CNV-seq), with mate pair sequencing used to confirm selected CNV deletion breakpoints. A total of 10 ng of input DNA was sufficient for accurate CNV-seq diagnosis, although 50 ng was optimal. Validation studies of samples with small CNVs showed that CNV-seq was specific and reproducible, suggesting that CNV-seq may have a potential genome resolution of approximately 0.1 Mb. In a blinded study of 72 samples with known gross and submicroscopic CNVs originally detected by single nucleotide polymorphism array, there was high diagnostic concordance with CNV-seq. We conclude that CNV-seq is a viable alternative to arrays for the diagnosis of chromosome disease syndromes. Detection of chromosome copy number variation (CNV) plays an important role in the diagnosis of patients with unexplained clinical symptoms and for the identification of chromosome disease syndromes in the established fetus. In current clinical practice, karyotyping, in conjunction with array-based methods, is the gold standard for detection of CNV. To increase accessibility and reduce patient costs for diagnostic CNV tests, we speculated that next-generation sequencing methods could provide a similar degree of sensitivity and specificity as commercial arrays. CNV in patient samples was assessed on a medium-density single nucleotide polymorphism array and by low-coverage massively parallel CNV sequencing (CNV-seq), with mate pair sequencing used to confirm selected CNV deletion breakpoints. A total of 10 ng of input DNA was sufficient for accurate CNV-seq diagnosis, although 50 ng was optimal. Validation studies of samples with small CNVs showed that CNV-seq was specific and reproducible, suggesting that CNV-seq may have a potential genome resolution of approximately 0.1 Mb. In a blinded study of 72 samples with known gross and submicroscopic CNVs originally detected by single nucleotide polymorphism array, there was high diagnostic concordance with CNV-seq. We conclude that CNV-seq is a viable alternative to arrays for the diagnosis of chromosome disease syndromes. There have been >200 chromosome disease syndromes recorded and studied in the human population.1Gardner R.J.M. Sutherland G.R. Shaffer L.G. Chromosome Abnormalities and Genetic Counselling. Oxford University Press, New York2011Crossref Google Scholar Whole or partial chromosome numerical changes account for most chromosome syndromes, with Down syndrome being the most prevalent. The remaining chromosome diseases are classified as submicroscopic deletion and duplication syndromes caused by the loss or gain of variably sized chromosome segments [copy number variations (CNVs)] that can affect the normal expression pattern of one or more genes. There is a wide spectrum of clinical phenotypes associated with chromosome disease syndromes, which can be categorized into multiple congenital abnormalities, physical disabilities, dysmorphic features, developmental delay, intellectual disability, seizure disorders, autistic behaviors, and learning disabilities.1Gardner R.J.M. Sutherland G.R. Shaffer L.G. Chromosome Abnormalities and Genetic Counselling. Oxford University Press, New York2011Crossref Google Scholar, 2Breman A. Pursley A.N. Hixson P. Bi W. Ward P. Bacino C.A. Shaw C. Lupski J.R. Beaudet A. Patel A. Cheung S.W. Van den Veyver I. Prenatal chromosomal microarray analysis in a diagnostic laboratory: experience with >1000 cases and review of the literature.Prenat Diagn. 2012; 32: 351-361Crossref PubMed Scopus (99) Google Scholar, 3Stankiewicz P. Beaudet A.L. Use of array CGH in the evaluation of dysmorphology, malformations, developmental delay, and idiopathic mental retardation.Curr Opin Genet Dev. 2007; 17: 182-192Crossref PubMed Scopus (264) Google Scholar Recent molecular analyses of human gametes and preimplantation human embryos4Nagaoka S.I. Hassold T.J. Hunt P.A. Human aneuploidy: mechanisms and new insights into an age-old problem.Nat Rev Genet. 2012; 13: 493-504Crossref PubMed Scopus (614) Google Scholar, 5Voet T. Vanneste E. Vermeesch J.R. The human cleavage stage embryo is a cradle of chromosomal rearrangements.Cytogenet Genome Res. 2011; 133: 160-168Crossref PubMed Scopus (45) Google Scholar, 6Vanneste E. Voet T. Le Caignec C. Ampe M. Konings P. Melotte C. Debrock S. Amyere M. Vikkula M. Schuit F. Fryns J.P. Verbeke G. D'Hooghe T. Moreau Y. Vermeesch J.R. Chromosome instability is common in human cleavage-stage embryos.Nat Med. 2009; 15: 577-583Crossref PubMed Scopus (619) Google Scholar have revealed that inherent chromosome instability is a major source of these chromosome abnormalities. Driven by well-known mechanisms of nonhomologous end joining, as well as some recently proposed mechanisms involving perturbation of DNA replication and replication of noncontiguous DNA segments, de novo CNVs form at rates far exceeding other kinds of mutagenic events.7Hastings P.J. Lupski J.R. Rosenberg S.M. Ira G. Mechanisms of changes in gene copy number.Nat Rev Genet. 2009; 10: 551-564Crossref PubMed Scopus (850) Google Scholar In addition, a small but significant proportion of chromosome disease continues to be perpetuated by inadvertent familial inheritance.1Gardner R.J.M. Sutherland G.R. Shaffer L.G. Chromosome Abnormalities and Genetic Counselling. Oxford University Press, New York2011Crossref Google Scholar Most of these chromosome abnormalities of de novo or parental origin are generally compatible with fetal development to term, leading to approximately 0.3% of children born with a chromosome disease.8Hassold T. Hunt P. To err (meiotically) is human: the genesis of human aneuploidy.Nat Rev Genet. 2001; 2: 280-291Crossref PubMed Scopus (1769) Google Scholar For >30 years, prenatal diagnosis has played a key role in detecting these chromosome abnormalities in the late first trimester or early second trimester of pregnancy, with the aim of reducing the incidence of chromosome disease in the newborn by elective termination of pregnancy.1Gardner R.J.M. Sutherland G.R. Shaffer L.G. Chromosome Abnormalities and Genetic Counselling. Oxford University Press, New York2011Crossref Google Scholar The application of prenatal diagnosis involves a range of fetus analyses, including screening tests, such as maternal serum screening and ultrasound, and if chromosome disease is suspected, follow-up diagnosis is usually performed by either chorionic villous sampling or amniocentesis and then karyotyping. Karyotyping, with a resolution of approximately five megabases (Mb), is the gold standard enabling the detection of fetal aneuploidies, polyploidies, balanced and unbalanced structural rearrangements, large microdeletions and duplications, and mosaicism when >20 metaphase cells are analyzed.9Bayani J. Squire J.A. Traditional banding of chromosomes for cytogenetic analysis.Curr Protoc Cell Biol. 2004; (Chapter 22:Unit 22)Google Scholar, 10Smeets D.F. Historical prospective of human cytogenetics: from microscope to microarray.Clin Biochem. 2004; 37: 439-446Crossref PubMed Scopus (64) Google Scholar, 11Trask B.J. Human cytogenetics: 46 chromosomes, 46 years and counting.Nat Rev Genet. 2002; 3: 769-778Crossref PubMed Scopus (161) Google Scholar Other methods, such as fluorescence in situ hybridization,12Jia C.W. Wang S.Y. Ma Y.M. Lan Y.L. Si Y.M. Yu L. Zhou L.Y. Fluorescence in situ hybridization in uncultured amniocytes for detection of aneuploidy in 4210 prenatal cases.Chin Med J (Engl). 2011; 124: 1164-1168PubMed Google Scholar quantitative fluorescent PCR,13Mann K. Hills A. Donaghue C. Thomas H. Ogilvie C.M. Quantitative fluorescence PCR analysis of >40,000 prenatal samples for the rapid diagnosis of trisomies 13, 18 and 21 and monosomy X.Prenat Diagn. 2012; 32: 1197-1204Crossref PubMed Scopus (25) Google Scholar and multiplex ligation-dependent probe amplification,14Van Opstal D. Boter M. de Jong D. van den Berg C. Bruggenwirth H.T. Wildschut H.I. de Klein A. Galjaard R.J. Rapid aneuploidy detection with multiplex ligation-dependent probe amplification: a prospective study of 4000 amniotic fluid samples.Eur J Hum Genet. 2009; 17: 112-121Crossref PubMed Scopus (61) Google Scholar have also been used on uncultured fetal cells to provide more rapid diagnosis of fetal aneuploidies. More recently, the introduction of high-resolution oligonucleotide and single nucleotide polymorphism (SNP) arrays has revolutionized prenatal diagnosis,15Shaffer L.G. Rosenfeld J.A. Microarray-based prenatal diagnosis for the identification of fetal chromosome abnormalities.Expert Rev Mol Diagn. 2013; 13: 601-611Crossref PubMed Scopus (16) Google Scholar enabling more comprehensive and accurate analysis of a wider spectrum of chromosome abnormalities, including clinically significant chromosome deletions and duplications <5 Mb in size.16Schaaf C.P. Wiszniewska J. Beaudet A.L. Copy number and SNP arrays in clinical diagnostics.Annu Rev Genomics Hum Genet. 2011; 12: 25-51Crossref PubMed Scopus (127) Google Scholar Current arrays in clinical application are generally custom designed using a high density of oligonucleotide or SNP probes uniformly covering the backbone of each chromosome as well as exonic regions of disease genes.3Stankiewicz P. Beaudet A.L. Use of array CGH in the evaluation of dysmorphology, malformations, developmental delay, and idiopathic mental retardation.Curr Opin Genet Dev. 2007; 17: 182-192Crossref PubMed Scopus (264) Google Scholar, 16Schaaf C.P. Wiszniewska J. Beaudet A.L. Copy number and SNP arrays in clinical diagnostics.Annu Rev Genomics Hum Genet. 2011; 12: 25-51Crossref PubMed Scopus (127) Google Scholar, 17Keren B. Le Caignec C. Oligonucleotide microarrays in constitutional genetic diagnosis.Expert Rev Mol Diagn. 2011; 11: 521-532Crossref PubMed Scopus (15) Google Scholar Dual oligonucleotide and SNP combination arrays have been designed for very high-resolution chromosome analysis and can detect additional clinically significant abnormalities missed by oligonucleotide arrays.18Fan Y.S. Ouyang X. Peng J. Sacharow S. Tekin M. Barbouth D. Bodamer O. Yusupov R. Navarrete C. Heller A.H. Pena SDj Frequent detection of parental consanguinity in children with developmental disorders by a combined CGH and SNP microarray.Mol Cytogenet. 2013; 6: 38Crossref PubMed Scopus (21) Google Scholar, 19Wiszniewska J. Bi W. Shaw C. Stankiewicz P. Kang S.H. Pursley A.N. Lalani S. Hixson P. Gambin T. Tsai C.H. Bock H.G. Descartes M. Probst F.J. Scaglia F. Beaudet A.L. Lupski J.R. Eng C. Wai Cheung S. Bacino C. Patel A. Combined array CGH plus SNP genome analyses in a single assay for optimized clinical testing.Eur J Hum Genet. 2014; 22: 79-87Crossref PubMed Scopus (97) Google Scholar Furthermore, with the availability of more detailed and comprehensive public databases of chromosome abnormalities and their associated clinical phenotypes, CNVs called by custom arrays can be readily translated into an accurate clinical diagnosis20Howard H.J. Beaudet A. Gil-da-Silva Lopes V. Lyne M. Suthers G. Van den Akker P. Wertheim-Tysarowska K. Willems P. Macrae F. Disease-specific databases: why we need them and some recommendations from the Human Variome Project Meeting, May 28, 2011.Am J Med Genet A. 2012; 158A: 2763-2766Crossref PubMed Scopus (11) Google Scholar and, in addition, has led to further knowledge on the genetic basis of more complex syndromes, such as autism.21Celestino-Soper P.B. Shaw C.A. Sanders S.J. Li J. Murtha M.T. Ercan-Sencicek A.G. Davis L. Thomson S. Gambin T. Chinault A.C. Ou Z. German J.R. Milosavljevic A. Sutcliffe J.S. Cook Jr., E.H. Stankiewicz P. State M.W. Beaudet A.L. Use of array CGH to detect exonic copy number variants throughout the genome in autism families detects a novel deletion in TMLHE.Hum Mol Genet. 2011; 20: 4360-4370Crossref PubMed Scopus (83) Google Scholar However, occasionally, some rare CNVs not represented in databases still remain problematic to give a firm diagnosis.22Klopocki E. Mundlos S. Copy-number variations, noncoding sequences, and human phenotypes.Annu Rev Genomics Hum Genet. 2011; 12: 53-72Crossref PubMed Scopus (48) Google Scholar Current data suggest that arrays can detect an additional 5% to 15% of chromosome abnormalities missed by karyotyping.16Schaaf C.P. Wiszniewska J. Beaudet A.L. Copy number and SNP arrays in clinical diagnostics.Annu Rev Genomics Hum Genet. 2011; 12: 25-51Crossref PubMed Scopus (127) Google Scholar, 23Wapner R.J. Martin C.L. Levy B. Ballif B.C. Eng C.M. Zachary J.M. Savage M. Platt L.D. Saltzman D. Grobman W.A. Klugman S. Scholl T. Simpson J.L. McCall K. Aggarwal V.S. Bunke B. Nahum O. Patel A. Lamb A.N. Thom E.A. Beaudet A.L. Ledbetter D.H. Shaffer L.G. Jackson L. Chromosomal microarray versus karyotyping for prenatal diagnosis.N Engl J Med. 2012; 367: 2175-2184Crossref PubMed Scopus (913) Google Scholar On the other hand, array-based diagnosis of polyploidies and balanced translocations still remains problematic. Based on a study of low- and high-risk pregnancies, it has been suggested that arrays could be used as a primary test to detect fetal chromosome abnormalities.24Fiorentino F. Napoletano S. Caiazzo F. Sessa M. Bono S. Spizzichino L. Gordon A. Nuccitelli A. Rizzo G. Baldi M. Chromosomal microarray analysis as a first-line test in pregnancies with a priori low risk for the detection of submicroscopic chromosomal abnormalities.Eur J Hum Genet. 2013; 21: 725-730Crossref PubMed Scopus (76) Google Scholar The combined application of ultrasound, karyotyping, and array technologies, particularly in Western countries, has had a substantial effect on our ability to identify fetal abnormalities in pregnancy, to provide an explanation for recurrent miscarriage, and to give a diagnosis to children and adults who experience unknown physical and mental problems, enabling better treatment regimens to improve their quality of life.16Schaaf C.P. Wiszniewska J. Beaudet A.L. Copy number and SNP arrays in clinical diagnostics.Annu Rev Genomics Hum Genet. 2011; 12: 25-51Crossref PubMed Scopus (127) Google Scholar, 25Callaway J.L. Shaffer L.G. Chitty L.S. Rosenfeld J.A. Crolla J.A. The clinical utility of microarray technologies applied to prenatal cytogenetics in the presence of a normal conventional karyotype: a review of the literature.Prenat Diagn. 2013; 33: 1119-1123Crossref PubMed Scopus (121) Google Scholar, 26Schwartz S. Clinical utility of single nucleotide polymorphism arrays.Clin Lab Med. 2011; 31: 581-594Abstract Full Text Full Text PDF PubMed Scopus (18) Google Scholar Nonetheless, in developing countries and some developed countries, lack of accessibility, inadequate expertise, and high cost have precluded the widespread use of arrays, and, therefore, the incidence of children born with chromosome disease remains unacceptably high.27Bian X.M. Guo Q. Qi Q.W. Current situation and development of prenatal diagnosis in China.Front Med China. 2010; 4: 271-274Crossref PubMed Scopus (7) Google Scholar, 28Kaur A. Singh J.R. Chromosomal abnormalities: genetic disease burden in India.Int J Hum Genet. 2010; 10: 1-14Google Scholar, 29Verma I.C. Burden of genetic disorders in India.Indian J Pediatr. 2000; 67: 893-898Crossref PubMed Scopus (91) Google Scholar Therefore, there is an unmet clinical need for an alternative technology that is equally comprehensive and accurate as arrays for most chromosome diseases but more affordable to all patients. We, therefore, speculated that a next-generation sequencing–based technology would fulfill this need. We previously showed that a low-coverage shotgun sequencing method of approximately five million mapped sequencing reads allocated to sequential 20-kb sequencing bins across each chromosome can detect levels of mosaicism of the X chromosome down to 5%.30Wang Y. Chen Y. Tian F. Zhang J. Song Z. Wu Y. Han X. Hu W. Ma D. Cram D. Cheng W. Maternal mosaicism is a significant contributor to discordant sex chromosomal aneuploidies associated with noninvasive prenatal testing.Clin Chem. 2014; 60: 251-259Crossref PubMed Scopus (203) Google Scholar We, therefore, hypothesized that this method could be equally applied to detect CNV at a relatively high resolution across the 22 autosomal chromosome pairs plus the sex chromosomes X and Y. Herein, we applied CNV sequencing (CNV-seq) to blinded DNA samples with known abnormalities defined by a medium-density SNP array and showed that this method was highly concordant, reproducible, and sensitive, with a potential resolution of approximately 0.1 Mb. Karyotyping and SNP array analyses of patient samples were performed at the State Key Laboratory of Medical Genetics, Central South University (Hunan, China), and Hunan Jiahui Genetics Hospital (Hunan, China). Genomic DNA samples selected for the study were from 62 patients with either developmental delay/intellectual disability or congenital abnormalities. In addition, genomic DNA from 10 miscarriage samples were also included in the study. Details of these 72 samples are shown in Supplemental Table S1. Additional research samples for further assessment of the sensitivity and specificity of CNV-seq included three samples with ring chromosomes and four samples with small CNVs <0.25 Mb. RNA-free high-molecular-weight DNA was prepared from patient blood and miscarriage tissue using the DNeasy blood and tissue kit (Qiagen GmbH, Hilden, Germany). The quality and concentration of genomic DNA samples was assessed by agarose gel electrophoresis using a NanoDrop spectrophotometer (Thermo Fisher Scientific, Wilmington, DE). Chromosome CNV analysis was performed using the HumanCytoSNP-12 BeadChip array (Illumina Inc., San Diego, CA), with an SNP probe density of 298,563 and average genome spacing of 19 kb. The log R ratio and the A and B allele frequency values were calculated using GenomeStudio software version 2011.1 (Illumina Inc.), and then detailed CNV analysis was performed using cnvPartition plug-in v3.1.6 software (Illumina Inc.). Duplications (AAA, AAB, ABB, BBB allelic combinations) and deletions (A or B allele) were defined by a confidence score >100 over a region of 50 SNPs. Fifity nanograms of genomic DNA was fragmented to an average size of 300 bp, and sequencing libraries were prepared as previously described.31Liang D. Lv W. Wang H. Xu L. Liu J. Li H. Hu L. Peng Y. Wu L. Non-invasive prenatal testing of fetal whole chromosome aneuploidy by massively parallel sequencing.Prenat Diagn. 2013; 33: 409-415Crossref PubMed Scopus (128) Google Scholar, 32Song Y. Liu C. Qi H. Zhang Y. Bian X. Liu J. Noninvasive prenatal testing of fetal aneuploidies by massively parallel sequencing in a prospective Chinese population.Prenat Diagn. 2013; 33: 700-706Crossref PubMed Scopus (160) Google Scholar Libraries were sequenced using the HiSeq 2000 platform (Illumina Inc.) to generate approximately 8 million 36-bp single-end reads, representing 0.1-fold genome coverage. All the sequences were aligned to the unmasked hg19 genome using the Burrows-Wheeler algorithm.33Li H. Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform.Bioinformatics. 2009; 25: 1754-1760Crossref PubMed Scopus (26904) Google Scholar A minimum of 20 batched test samples were internally compared with each other as reference using the data processing and analysis algorithms as previously described.30Wang Y. Chen Y. Tian F. Zhang J. Song Z. Wu Y. Han X. Hu W. Ma D. Cram D. Cheng W. Maternal mosaicism is a significant contributor to discordant sex chromosomal aneuploidies associated with noninvasive prenatal testing.Clin Chem. 2014; 60: 251-259Crossref PubMed Scopus (203) Google Scholar To improve detection sensitivity, a 60-kb window with 5-kb sliding was finally used to allocate and analyze approximately 5 million sequencing reads in overlapping 60-kb bins. For each chromosome, we plotted the mean log2 values of the normalized sequencing reads (y axis) versus the number of sequential 5-kb sliding 60-kb sequencing bins (x axis). The mean log2 value was then calculated along the length of each chromosome. The theoretical log2 value for a duplication (three copies) is log2 [1.5] = 0.58 and for a deletion (one copy) is log2 [0.5] = −1.0. Cutoff copy number values used to call duplications were set at >2.8 (log2 [1.4] = 0.49), and those used to call deletions were set at <1.2 (log2 [0.6] = −0.74). Mate pair sequencing34Asan Geng C. Chen Y. Wu K. Cai Q. Wang Y. Lang Y. Cao H. Yang H. Wang J. Zhang X. Paired-end sequencing of long-range DNA fragments for de novo assembly of large, complex Mammalian genomes by direct intra-molecule ligation.PLoS One. 2012; 7: e46211Crossref PubMed Scopus (8) Google Scholar was performed using the Nextera mate pair sample preparation kit (Illumina Inc.). After fragmentation of 5 μg of genomic DNA, resulting fragments were resolved on 1% agarose gels, and the 5-kb fraction was gel purified using the QIAquick gel extraction kit (Qiagen GmbH). Libraries were sequenced using the HiSeq 2000 platform to produce paired forward and reverse end reads of approximately 100 bp. Sequence reads were then uniquely mapped to the hg19 reference genome using the Burrows-Wheeler algorithm.33Li H. Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform.Bioinformatics. 2009; 25: 1754-1760Crossref PubMed Scopus (26904) Google Scholar A total of 10 to 25 million sequencing reads were generated to identify at least one paired read spanning the deletion breakpoint. We previously reported stable and accurate CNV-seq chromosome profiles consistent with known karyotypes of genomic DNA samples, where the initial input DNA amount for library construction was 100 ng.30Wang Y. Chen Y. Tian F. Zhang J. Song Z. Wu Y. Han X. Hu W. Ma D. Cram D. Cheng W. Maternal mosaicism is a significant contributor to discordant sex chromosomal aneuploidies associated with noninvasive prenatal testing.Clin Chem. 2014; 60: 251-259Crossref PubMed Scopus (203) Google Scholar Given the variability in the yield of genomic DNA isolated from clinical samples, the clinical utility of this CNV-seq method was first tested on low template DNA extracted from a normal sample (46,XX) and an abnormal sample from a patient with Wolf-Hirschhorn syndrome (46,XX,4p16.1-pter; 8.92-Mb deletion). Lower input DNA amounts of 50 and 10 ng were prepared in duplicate and were subjected to CNV-seq (Figure 1, 50- and 10-ng duplicate 46,XX and 46,XX,4p16.1-pter CNV-seq profiles of chromosome 4; Supplemental Figures S1 and S2, one 50-ng replicate 46,XX,4p16.1-pter CNV-seq profile of all chromosomes). Normal chromosome profiles exhibiting no significant CNV were observed for 50-ng 46,XX and 46,XX,4p16.1-pter samples, except in the latter, the expected full copy number loss was observed for the 4p16.1-pter region. In contrast, chromosome CNV profiles were slightly less stable in the 10-ng normal and abnormal samples, with some chromosomes showing very minor nonspecific CNV fluctuations, particularly at the terminal regions, and in the abnormal samples, slightly less than a full copy number loss of 4p16.1-pter was observed. On the basis of these initial investigations, 50 ng of genomic DNA seemed to be more optimal for reliable CNV analysis, at an 8 million read sequencing depth. From analysis of research genomic DNA samples, we occasionally detected deletions and duplications a few hundred kilobases in size. A natural question to ask was whether these small CNVs were in fact real. To address the specificity of CNV-seq for detecting true copy number changes, we used mate pair sequencing to confirm and precisely define the deletion interval and the breakpoint fusion sequences in two genomic DNA research samples originally analyzed by CNV-seq. These samples included one harboring an apparent 0.22-Mb Xp22.2 deletion, which we suspected resulted in the loss of the PLP1 gene and X-linked Pelizaeus-Merzbacher disease, and the other an apparent 0.08-Mb homozygous 6q26 deletion, which we suspected was associated with loss of the Parkinson disease gene PARK2. In both DNA samples (Figures 2 and 3), at least two 5-kb DNA fragments spanning the deletion breakpoints were identified by mate pair sequencing. Based on the paired-end sequencing results, we designed PCR primers and amplified smaller fragments encompassing the two breakpoints. Sanger sequencing pinpointed the breakpoints down to the single nucleotide level in the suspected genes, confirming that both deletions originally detected by CNV-seq were, indeed, bona fide deletions.Figure 3Reproduciblity of CNV-seq for CNV. Triplicate CNV profiles showing the 0.22-Mb Xp22.2 and 0.08-Mb 6q26 deletions. CNV-seq profiles are log2 values of the normalized sequencing read densities (y axis) versus the number of sequential 5-kb sliding 60-kb sequencing bins (x axis). The upper [log2 (3/2)] and lower [log2 (1/2)] dashed lines indicate 100% (one copy) chromosome gain (duplication) and loss (deletion), respectively. The mean log2 value for CNV (blue tracking line), regions of repetitive sequences (red bars), and the centromere (black bars) are also shown. Both deletions (arrows) were reproducibly detected by CNV-seq in three independent replicate samples and showed the expected copy number change.View Large Image Figure ViewerDownload Hi-res image Download (PPT) To determine whether we could reproducibly detect these two small CNVs, libraries were constructed from triplicate 50-ng genomic DNA samples and were reanalyzed by CNV-seq. The 0.22-Mb Xp22.2 deletion and the 0.08-Mb homozygous 6q26 deletion, demonstrated previously herein to be bona fide deletions by mate pair sequencing, were reproducibly detected in all three replicate samples and consistently showed only the expected single and double copy changes, respectively (Figure 3). Furthermore, two other genomic DNA samples were subsequently analyzed, the first exhibiting a 0.24-Mb 22q11.2 deletion causing hemizygous expression of three nondisease Online Mendelian Inheritance of Man genes (IGLL3P, LRP5L, and CRYBB2P1) and the second exhibiting a 0.22-Mb 9q33.1 deletion causing hemizygous expression of the disease Online Mendelian Inheritance of Man gene TRIM32. The expected single copy number change, at the expected location only, was reproducibly observed in both samples (data not shown). On the basis of these findings, we concluded that CNV-seq was highly specific and reproducible for detecting these small CNVs. The suitability of CNV-seq for comprehensive diagnosis of chromosome disease syndromes was assessed on 72 samples with gross chromosome and submicroscopic chromosome abnormalities that were previously diagnosed by the medium-density HumanCytoSNP-12 BeadChip array. Genomic DNA from the 72 samples, some of which had been stored for more than 5 years, were coded, and blinded samples were sent for independent CNV-seq analysis. After CNV-seq analyses, samples were unblinded and diagnoses were compared with the original SNP diagnoses (Table 1, data summary; Supplemental Table S1, complete data). Eleven selected comparative SNP and CNV-seq profiles are shown as examples; Figure 4 shows four chromosome syndromes, and Supplemental Figure S3 shows seven additional chromosome syndromes. For the 72 samples, diagnosis of the primary chromosome abnormality was fully concordant by SNP array and CNV-seq (Table 1 and Supplemental Table S1). In 43 of of these samples (60%), CNVs could be clearly associated with known chromosome disease syndromes. In addition, CNV-seq detected five secondary CNVs of <1 Mb (0.20 to 0.66 Mb) of unknown clinical significance that were not originally detected by the SNP array. In all concordant samples, the nature and size of the gross submicroscopic chromosome deletion or duplication events, and the location of the anomaly was almost identical or very similar between SNP array and CNV-seq, and diagnoses were generally consistent with karyotyping data. Taken together, these validation studies performed on blinded samples indicate that CNV-seq can provide a reliable and accurate diagnosis of a range of different chromosome aberrations and can perform similarly to a medium-density SNP array.Table 1Diagnostic Concordance of SNP Array and CNV-SeqSample typeNo. of samplesSNP/CNV-seq [No./total No. (%)]Primary CNVSecondary CNVAll CNVsPatient blood6262/62 (100)11/16 (69)∗All five discordant results were due to the inability of the HumanCytoSNP-12 BeadChip array to detect the CNV with a confidence score >100.73/78 (93)∗All five discordant results were due to the inability of the HumanCytoSNP-12 BeadChip array to detect the CNV with a confidence score >100.Miscarriage1010/10 (100)1/1 (100)11/11 (100)Total7272/72 (100)12/17 (71)∗All five discordant results were due to the inability of the HumanCytoSNP-12 BeadChip array to detect the CNV with a confidence score >100.84/89 (94)∗All five discordant results were due to the inability of the HumanCytoSNP-12 BeadChip array to detect the CNV with a confidence score >100.∗ All five discordant results were due to the inability of the HumanCytoSNP-12